Tube Power Amp IRs and Captures

[djd100]

If you’re implying that anything above 8k doesn’t matter because it can’t be heard, then I flatly disagree. I’m often eq-ing that area on guitars and even subtle amounts can be noticeable.



AS YOU WISH:

View attachment 38915View attachment 38916

Others have demonstrated this and it’s also audible. I’m not saying it is an issue but it’s easy to demonstrate that it goes on.

Well, the quiet take looks worse but who knows how the input levels were set, and the loud take looks much closer, so it seems ToneX is more input sensitive, but with a median absolute error delta of roughly 2db, and the ToneX capture roll-off starting at 15K, I wouldn't consider ToneX to be unusable for a tube power amp sim that runs through guitar speakers or IR's that roll-off much earlier, noting again that not all tube guitar preamps are heavily filtered (guitar preamp filtering typically takes place on the bottom end, excepting some higher gain preamps which filter highs for fizz control, and that many guitar preamps are 20Hz to 20Khz).

Also note that ToneX includes an EQ, so upping the high end a tad is certainly doable for those who might need it.

So again, and for the last time, for the purpose of a tube power amp sim or a full amp sim, while NAM is of course better all around no doubt (never claimed it wasn't!), ToneX is perfectly usable for the intended purpose of a tube power amp sim or full amp sim running through the additional LP/HP filtering provided by guitar speakers and cab IR's.

Hopefully the ToneX software can be improved going forward, but if it isn't it still works fine in the real world, and it's the only capture solution that can be currently had in the Hammond 1590A enclosure size (the ToneX One, which is why I use it as it's small and fits on a pedalboard with my tube preamp(s) and a FM3).

Anyway, thank you for posting this, and to quote your source, "I’m not saying it is an issue...".

 

[KingsXJJ]

I’m using it to make a (made up full scope signal chain) to sound like pedals->amp->cab->effects->studio mic sound or pedals->amp->fx loop->amp->cab “in the room”. Pedals and effects can be whatever you want where you want.

Somebody screenshot that shit and send it to 4chan.

 

[Progressive_disclosure]

This is a much deeper topic than I thought!

If I can manage I’ll try make my own little test of this. Mind that I’m not looking to purchase any additional hardware now, just understand how well different available solutions would work for my use case. I have an amp with a built-in resistive load and my purpose of simulating the power amp is to see if I’m missing anything when silent recording the whole amp via the line out versus recording just the preamp and simulating both power amp and speaker.

I’ll test making power amp models of just one setting on the amp with both NAM and Tonex. Signal chain would be:

HX Stomp FX loop out (instrument level) -> FX return on amp -> line out of amp -> UA Volt 2 input (line level). I’ll use my 1x12 V30 as speaker for the captures. Don’t see a point in capturing using the resistive load.

Then I’ll record some DI’s and send through the preamp section only first and then through both preamp and power amp twice (once with the resistive load and once without). For the preamp only I’ll set separate tracks using the NAM and Tonex power amp captures.

Summary I’m hoping for is four tracks:
Preamp - Tonex - IR
Preamp - NAM - IR
Whole amp w resistive load - IR
Whole amp w speaker load - IR

Don’t know if I’ll have the time right now to do all this but after it’s done I can hopefully build my own impression of how I want to proceed. I don’t want to go into IRs vs real speakers as @GuitarJon has that covered. I’ll just use the same IR for everything. I’ll use the same settings on the amp throughout all this. Will record both clean and hi-gain DI’s.

Anything I should tweak in the setup above? I have no idea how to get a frequency response visualization. Does anyone know an easy way of getting that in Logic?

 

[MirrorProfiles]

Well, the quiet take looks worse but who knows how the input levels were set

He literally explains in the post how they were set, and even tweaked them to give ToneX the most accurate result (which shouldn’t be required, but he didn’t want the test to seem skewed in favour of NAM).

the ToneX capture roll-off starting at 15K,

It starts rolling off around 6kHz. By 15kHz it’s way off the source.

I wouldn't consider ToneX to be unusable for a tube power amp sim that runs through guitar speakers

There is a difference between unusable and inaccurate. Whether someone likes the results or not has nothing to do with accuracy, which is kind of why these graphs are helpful in the first place.

not all tube guitar preamps are heavily filtered (guitar preamp filtering typically takes place on the bottom end, excepting some higher gain preamps which filter highs for fizz control, and that many guitar preamps are 20Hz to 20Khz).

So is accuracy at 6kHz and above important or not? IMO it is, our ears are very sensitive in that area and it’s not a part of the signal I really want to be off. Above 12k or so the signal is reduced a lot already and difficult to hear. From 6kHz to 10k it’s pretty apparent.

Also note that ToneX includes an EQ, so upping the high end a tad is certainly doable for those who might need it.

An accurate poweramp model wouldn’t need this, though. Fine for a botch fix but not as something that is required, IMO.

ToneX is perfectly usable for the intended purpose of a tube power amp sim or full amp sim running through the additional LP/HP filtering provided by guitar speakers and cab IR's.

So despite the EXACT evidence you asked for being posted for you showing it, you’ve just decided it doesn’t matter anyway. And even though there is plenty going on above 6k in a signal, you’ve also concluded that it doesn’t matter.

The point being made was that ToneX cannot reproduce above 6k or so with as much accuracy as it can below that. Whether you like the results regardless is up to you. Anyone can do the same test for themselves and get the same result, it’s a proven fact at this point.

 

[AFKAEjay(retired)]

There is a difference between unusable and inaccurate.

In my mind “inaccurate” gets 2 subdivisions: relevant / irrelevant for the usecase.

So despite the EXACT evidence you asked for being posted for you showing it, you’ve just decided it doesn’t matter anyway.

I’d also confidently use tonex as a poweramp thingy based on these numbers tbh…inaccurate above 7k…while the signal goes into a guitar speaker that almost rolls off at 5k. Most speakers drop 25db above 5k? 6db reduction being 50%…not much left to be f upped by stuff above 7k.

Would I use tonex to capture a vocal preamp based on these numbers?…probably not, cause in my mind the very upper frequencies are relevant there.

 

[MirrorProfiles]

In my mind “inaccurate” gets 2 subdivisions: relevant / irrelevant for the usecase.

I’d also confidently use tonex as a poweramp thingy based on these numbers tbh…inaccurate above 7k…while the signal goes into a guitar speaker that almost rolls off at 5k. Most speakers drop 25db above 5k? 6db reduction being 50%…not much left to be f upped by stuff above 7k.

Would I use tonex to capture a vocal preamp based on these numbers?…probably not, cause in my mind the very upper frequencies are relevant there.

It’s not like it’s so rolled off that it’s negligible. Sweep around with an EQ and notice where you can start hearing the effect of it. You can hear even minor changes easily from 10k and below and it can have a drastic effect on the guitar sound. And if you use a LPF, you can hear the effects of it from around 12kHz (because it effects frequencies below that gradually). I'm doing EQ work on guitars on those areas all the time.

Above 12k, fair enough (although I'd still prefer to be the one to decide how the signal is processed). I’m only even aware of ToneX doing this because it is audible.

Shouldn't need a video to demonstrate but just to end this assumption that nothing matters above 6k. Even with a filter at 12k, you can hear the effects of it changing the guitar tone. IMO even losing that resolution at 12k would be a dealbreaker for me achieving accurate models. Why would I want to automatically lose accuracy in such a critical area without deciding it for myself?

 

[djd100]

He literally explains in the post how they were set, and even tweaked them to give ToneX the most accurate result (which shouldn’t be required, but he didn’t want the test to seem skewed in favour of NAM).

It starts rolling off around 6kHz. By 15kHz it’s way off the source.


There is a difference between unusable and inaccurate. Whether someone likes the results or not has nothing to do with accuracy, which is kind of why these graphs are helpful in the first place.


So is accuracy at 6kHz and above important or not? IMO it is, our ears are very sensitive in that area and it’s not a part of the signal I really want to be off. Above 12k or so the signal is reduced a lot already and difficult to hear. From 6kHz to 10k it’s pretty apparent.

An accurate poweramp model wouldn’t need this, though. Fine for a botch fix but not as something that is required, IMO.


So despite the EXACT evidence you asked for being posted for you showing it, you’ve just decided it doesn’t matter anyway. And even though there is plenty going on above 6k in a signal, you’ve also concluded that it doesn’t matter.

The point being made was that ToneX cannot reproduce above 6k or so with as much accuracy as it can below that. Whether you like the results regardless is up to you. Anyone can do the same test for themselves and get the same result, it’s a proven fact at this point.

The deviations start around 10K for the loud take, and around 7K for the quiet take (plus low end deviations below 50Hz), and they're only only a few db at that point, while nobody ever said ToneX was as accurate as NAM, though what was said is that you won't hear major differences in real world usage, especially in use with other instruments, i.e. a mix.

You can touch up the Tonex's top end with a little of it's onboard EQ if desired, and you'll be close enough for government work considering that the high mids and up stuff will already be rolled off via the cab or IR, and of course masked by other instruments (assuming you play with other musicians?).

I don't what you're arguing about as you make it sound like Tonex has no high end above 6K which is not true. So, to summerize again for everyone else, and for the last time, NAM is better, ToneX is good enough to get the job done, and ToneX is available in small inexpensive HW for end use along with some basic FX.

Cheers, I'm out...

 

[MirrorProfiles]

I don't what you're arguing about as you make it sound like Tonex has no high end above 6K which is not true.

No. I am saying ToneX lacks accuracy above 6k, not that it lacks high end. I am saying the information above 6k is important, and even small variances can change the sound of an amp. I just showed how much different anything above 6k makes to the tone, so it needs to be accurate. Rolling off sounds very different.

Check what @Slammin Mofo said - we have both made many models on many platforms and have each done lots of comparisons. So have many others.

I've measured and compared quite a few poweramp settings of my Powerstation and Mesa 2:90 poweramps and the Tonex models do not sound accurate compared to a real poweramp. The difference is very noticeable, much darker tone and less dynamic response on the Tonex model compared to the NAM model, which is very close to the real thing, and the real poweramps.

There is no dispute about the accuracy (its easily provable) - the only discussion is that you are intent on telling me that it doesn't present a problem (to you, for however you are using it), and I am saying its enough of a problem to make me not make ToneX models, because the models do not sound accurate enough in the top end relative to how much time they take to train. Is there anything to disagree with there? It's not about whether ToneX is accurate in the top end, because that's demonstrable. It's just whether or not it's a problem and that's circumstantial.

 

[EOengineer]

TONEX IS THE BEST TECHNOLOGY TO EVER EXIST AND MY TONEX CAN BEAT UP YOUR DAD.

 

[AFKAEjay(retired)]

It’s not like it’s so rolled off that it’s negligible. Sweep around with an EQ and notice where you can start hearing the effect of it. You can hear even minor changes easily from 10k and below and it can have a drastic effect on the guitar sound. And if you use a LPF, you can hear the effects of it from around 12kHz (because it effects frequencies below that gradually). I'm doing EQ work on guitars on those areas all the time.

Above 12k, fair enough (although I'd still prefer to be the one to decide how the signal is processed). I’m only even aware of ToneX doing this because it is audible.

Shouldn't need a video to demonstrate but just to end this assumption that nothing matters above 6k. Even with a filter at 12k, you can hear the effects of it changing the guitar tone. IMO even losing that resolution at 12k would be a dealbreaker for me achieving accurate models. Why would I want to automatically lose accuracy in such a critical area without deciding it for myself?

I definitely don’t think above 6k doesn’t matter in the end result…no discussion there.
The point is was making that >7k inaccuracies into a device (cab) that rolls off at 5k doesn’t matter.

The vid shows more audible differences at 7kish then i would have expected through a guitar cab though. Is that IR respresenting a cab curve…or hyped in the highs to begin with?..or unknown?

Maybe a more insigntfull test would be what would happen if you apply the same EQ before the IR?
Now you add around 3 DB right?, 25% of volume in that frequency range.
If you do it before the IR, that same +3db gets cut rougly 20Db by the IR…cant get my head around the math tbh…is the impact of that 3db still the same?

 

[[Nathan]]

TONEX IS THE BEST TECHNOLOGY TO EVER EXIST AND MY TONEX CAN BEAT UP YOUR DAD.

 

[MirrorProfiles]

The vid shows more audible differences at 7kish then i would have expected through a guitar cab though. Is that IR respresenting a cab curve…or hyped in the highs to begin with?..or unknown?

It’s a real BE100 with a regular IR. No hyping of anything going on, as flat as you’ll get. Totally representative of any mic’d cabinet.

Maybe a more insigntfull test would be what would happen if you apply the same EQ before the IR?
Now you add around 3 DB right?, 25% of volume in that frequency range.
If you do it before the IR, that same +3db gets cut rougly 20Db by the IR…cant get my head around the math tbh…is the impact of that 3db still the same?

IR’s are LTI. You can EQ before or after and the result is the same.

I’m currently making a ToneX poweramp model - will this be enough to prove a point or are we just going to keep moving goalposts?

 

[EOengineer]

Im a moderately happy user of Tonex in a weekly working environment, but I did get the sense that I preferred the NAM captures I attempted a couple months back.

I definitely pick up the slightly attenuated top end in Tonex and how that impacts the liveliness of the tones. In particular high gain captures have been a challenge.

Nonetheless it’s good enough to be useful. I’ll likely switch over to NAM once there’s hardware that doesn’t look like the unibomber slapped it together in a shed.

 

[AFKAEjay(retired)]

will this be enough to prove a point or are we just going to keep moving goalposts?

You don’t have to prove anything my friend, I’m not even a Tonex or NAM user, I’m just a guy using preamp captures into poweramp captures with an option (or eq blocks to replace those cause poweramp captures sum to mono in the QC)

And goalpost sometimes move when thought processes develop…at least in my case :)

I would find it very interesting though to hear differences between preamp A going into a Tonex/nam poweramp captures. QC even more…cause that’s what I use.

I have on my personal to do list to compare real, to DI capture, to capture of preamp + DI capture of poweramp on the QC.
Should be the same…

Btw…what surprises me in the graphs you shared…that particular poweramp seems to boost lows, and roll off highs…while in my perception poweramps roll of low mids, and enhence présence. Whats your take on that observation?

 

[MirrorProfiles]

Btw…what surprises me in the graphs you shared…that particular poweramp seems to boost lows, and roll off highs…while in my perception poweramps roll of low mids, and enhence présence. Whats your take on that observation?

It really depends, the presence and resonance circuits are in the power amp so those settings will have a big effect on the power amp response. Different power amp circuits have different amounts of negative feedback too, so that can influence how much the cab affects things. Typically I think of power amps as adding bass and top end, but if you are slamming into one hard and distorting it, those parts of the frequency spectrum will hit the headroom limit first and it'll sound more compressed and midrange/thick by comparison. Its kind of why a static EQ can't approximate it

 

[djd100]

No. I am saying ToneX lacks accuracy above 6k, not that it lacks high end. I am saying the information above 6k is important, and even small variances can change the sound of an amp. I just showed how much different anything above 6k makes to the tone, so it needs to be accurate. Rolling off sounds very different.

Check what @Slammin Mofo said - we have both made many models on many platforms and have each done lots of comparisons. So have many others.



There is no dispute about the accuracy (its easily provable) - the only discussion is that you are intent on telling me that it doesn't present a problem (to you, for however you are using it), and I am saying its enough of a problem to make me not make ToneX models, because the models do not sound accurate enough in the top end relative to how much time they take to train. Is there anything to disagree with there? It's not about whether ToneX is accurate in the top end, because that's demonstrable. It's just whether or not it's a problem and that's circumstantial.

No. I am saying ToneX lacks accuracy above 6k, not that it lacks high end. I am saying the information above 6k is important, and even small variances can change the sound of an amp. I just showed how much different anything above 6k makes to the tone, so it needs to be accurate. Rolling off sounds very different.

Check what @Slammin Mofo said - we have both made many models on many platforms and have each done lots of comparisons. So have many others.



There is no dispute about the accuracy (its easily provable) - the only discussion is that you are intent on telling me that it doesn't present a problem (to you, for however you are using it), and I am saying its enough of a problem to make me not make ToneX models, because the models do not sound accurate enough in the top end relative to how much time they take to train. Is there anything to disagree with there? It's not about whether ToneX is accurate in the top end, because that's demonstrable. It's just whether or not it's a problem and that's circumstantial.

Ok, NOW HEAR THIS, all ToneX users get out your hammers and smash those pieces of crap and delete it from your computers, then be happy to know that you'll never have to attempt to listen for your worthless ToneX tube power amp captures being 28db down at 8k, rather than NAM's being only 25db down when used through guitar speakers or guitar cab/speaker IR's!

Whatever you worthless ToneX users do, do not even consider using the ToneX's EQ to add 3db @ 8k if you feel your fake tube power amp tone is not bright enough, because only NAM can make you a star and get you the hot babes!

Finally, don't let the NAMA's catch you using that old obsolete inaccurate ToneX crap, because they will post your eyes out showing you the error of your ways LOL!

 

[KingsXJJ]

Whatever you worthless ToneX users do, do not even consider using the ToneX's EQ to add 3db @ 8k if you feel your fake tube power amp tone is not bright enough, because only NAM can make you a star and get you the hot babes!

I like hot babes! If one only has ToneX, what sort of babes does one get? Asking for a friend.

 

[djd100]

I like hot babes! If one only has ToneX, what sort of babes does one get? Asking for a friend.

Bag babes...

 

[MirrorProfiles]

Ok, NOW HEAR THIS, all ToneX users get out your hammers and smash those pieces of crap and delete it from your computers, then be happy to know that you'll never have to attempt to listen for your worthless ToneX tube power amp captures being 28db down at 8k, rather than NAM's being only 25db down when used through guitar speakers or guitar cab/speaker IR's!

Whatever you worthless ToneX users do, do not even consider using the ToneX's EQ to add 3db @ 8k if you feel your fake tube power amp tone is not bright enough, because only NAM can make you a star and get you the hot babes!

Finally, don't let the NAMA's catch you using that old obsolete inaccurate ToneX crap, because they will post your eyes out showing you the error of your ways LOL!

Lordy. Did you get Black Angus’d today or what?

 

[EOengineer]

Wait I already smashed my Tonex did I misread the room?