Jump to content
amsdenj

IR referencing and hi cut

Recommended Posts

post-10580-0-74133600-1487466998_thumb.jpg
Did some referencing today after repairing these two old Fender amps. The Tremolux on the Left is our bass player's old guitar amp. It needed an new bias filter cap. The Showman on the right is my amp from 1968. It needed a new output transformer, new tubes, new filter caps in the power supply and a new optocoupler for the tremolo circuit. The Temolux has the original Oxford 10K5's while the Showman has a pair of new Celestion G12-65's. That's an old Lexicon MPXG2 guitar effects processor on top of the Showman. Its still pretty interesting, but not used in this test. 
 
While testing the amps I thought I might not only compare the Tremolux and Showman, but also use them to reference my typical Helix patch, just to make sure I'm in the ballpark. I typically use that JBL EON610 as my vocal and guitar monitor so I used that to compare against the reference.
 
The Tremolux and Showman certainly sound different, I think mostly because of the different speakers and the different tremolo circuit. The Tremolux is the older bias tremolo while the Showman is the newer optical tremolo. The difference is that the optical tremolo cuts some of the gain in the Normal channel, even if the tremolo is off. That doesn't happen with the bias tremolo. Although they sound different, they are clearly the very similar sounding amps. That Tremolux has a little magic going.
 
Here's the surprise. I have the Celestion G12-65 IRs, and one I made myself of that exact cabinet with the Showman. I found using my IR and a Deluxe Vib amp model, I got an almost exact replica of that Showman amp coming out of Helix and that EON610. However, to do so, I had to turn the high cut on the IR way down to about 3.2 KHz. The Celestion G12-65 and all the other IRs I had were even brighter.
 
I was very surprised that all the IRs seemed to be a lot brighter through that EON610 than either reference amp. I found that I had to turn the high cut down to between 2 and 3 KHz to match the tone, and to achieve a warm pleasant tone from the Helix.
 
Have others noticed that the IRs tend to be very bright and need a lot of high cut in Helix? The 3sigmaaudio IRs were especially bright. They sound great with a lot of high cut, but I don't understand why this would be needed. I don't recall having this problem using the same IRs in S-Gear, but that's though different monitor speakers.
 
It could be the EON610. The response curve doesn't look like there's a lot of high boost, and I have the EQ turned off in the speaker. But maybe that horn is adding a lot of hype to the high end.
 
So the lesson is referencing real amps and then adjusting Helix IR low and hi cut to compensate for your FRFR can be a valuable and surprising learning experience.
 
 

 

post-10580-0-74133600-1487466998_thumb.jpg

  • Upvote 5

Share this post


Link to post
Share on other sites

I appreciate your thorough post! I've found myself cutting down that low and questioned my sanity. There comes a point where, even though my ears tell me it's good, the number staring at me from the screen makes me question what I'm doing. It just seems odd. Glad to hear some corroboration from a clearly knowledgeable source! I'm using an Atomic CLR cab.

  • Upvote 2

Share this post


Link to post
Share on other sites

To my ear, the speaker/IR high cut seems not very steep, maybe 6db/octave. I think L6 has said that for the built in cabs, it behaves like the actual cab, not sure. But I've often wished for both steepness and resonance controls. My hope would be to do a better job of losing the inappropriate highs, while boosting a bit just below where that roll-off kicks in, to keep it from turning to mud.

  • Upvote 1

Share this post


Link to post
Share on other sites

yes! i've used plug in filters that do that in my DAW and they can sound pretty good on some things.. that little boost near the cut-off

Share this post


Link to post
Share on other sites

I think mic and mic placement also has a bit to do with how much you might need to cut, and it seems to vary a bit with the amp model as well.  There are some combinations of amp models and IRs that need little to no adjustment particularly when overdriven, but most seem to.

 

I've come to the conclusion this likely has more to do with the difference in the design of the speaker cabinets in that powered speakers like your JBL or my Yamaha DXR have a LOT more sound energy coming out of the front and almost no energy loss at the back of the cabinet.  I've noticed when I turn the speaker down, when I stand back away from the speaker, or when I capture it in a recording it's a lot less harsh on the high end, which seems to indicate that some of it is in the way we hear these speakers.

  • Upvote 1

Share this post


Link to post
Share on other sites

How did you reference and compare them? Was that by recording or by comparing amp out vs Helix out in the room?

I think he's just comparing plugging into the amp and playing and plugging into the Helix through his JBL EONs and playing.  At least that's how I've made my comparisons.  Although that's not actually a good direct comparison since going through the Helix and an IR means you're incorporating a mic along with the placement of the mic which you're not doing when going straight through the amp.  But there's no doubt there's more upper end harshness in many cases when going through the Helix and most IRs.

Share this post


Link to post
Share on other sites

I have spent a lot of time recording my own tube amps and cabs (blues deluxe, ac15htv, H&K, etc..) with several mics (tlm 103, beyerdynamic tg60, e609, sm57, etc) and I have to say Helix is pretty much the real thing, when compared to that kind of sound. And I never had an amp sounding the same in two different locations or positions, while Helix gives me more consistency.

Just my 2¢..

  • Upvote 1

Share this post


Link to post
Share on other sites

My goal with this reference exercise wasn't to make Helix sound just like the real amps, it was just to see if my gigging Helix tone through the EON610 monitor and FOH was doing what I expected or not. The motivation for this was last gig I switched from JOOSTALNICO-G12M-R121-U67 (which is quite warm) to the recent Celestion G12-65 Balanced IR. This sounded overly bright to me at the gig so I through maybe I should check against a reference to see if I was in the ballpark.

 

I'm happy enough to turn down the high cut to whatever is needed to make the 610 sound good. I'm also OK with that being something unique about that speaker, and maybe something I'll fix in the speaker itself (if I could ever get EON Connect to actually connect so I can adjust the EQ!).

 

I would just like to understand where all that brightness is coming from - the Helix amp models, the IRs, and/or the EON 610. I'm going to do some more testing today with a different FRFR so I can narrow down what's causing the overly bright tone. I'm suspecting that 610. If so, and I can get the EQ to work and roll off the highs in the 610, then that's fine.

Share this post


Link to post
Share on other sites

If you're comparing what the amps sound like compared to the JBL at the distance from your Helix in the picture, of course the IR is going to sound brighter. It's replicating a close-mic sound. Depending on the music and tones, a low pass that low in the highs might work fine, but it's probably not very accurate to what those amps sound like mic'ed through a PA. At the end of the day, though, it's all about getting the sound YOU want.

Share this post


Link to post
Share on other sites

It's not uncommon for me to turn the Hi Cut all the way down to 2kh.  I try not to worry about the numbers and tweak till it sounds good to me.

Share this post


Link to post
Share on other sites

Interesting subject; something I've wondered about.

 

Here's another experiment I did by the way;

 

I compared a cab block (separate to an amp block) with nothing else in the chain.

I did high and low cuts to taste in the cab block.

Then I'd also dropped a high/low-cut block right after the cab block. I duplicated the settings in both, and then I compared.

The cuts WHITHIN the cab block have FAR less effect on the results than the cuts following it.

 

I just realized; the other test I should've done to compare; put a duplicate high/low-cut block in between the amp and cab, just to see what that would do.

I suspect that would be more subtle than the other locations though, being as it would only impart its effects on the signal path up to and including the amp, master and all.

 

I found that, with the block FOLLOWING the cab, I had to drop it almost 3kHz further to get a similar result.

 

Sorry these numbers aren't super-detailed; I was doing this testing in a bit of a hurry this past Friday, with intention of revisiting the topic later.

 

This might inspire someone to do similar experiments.

Share this post


Link to post
Share on other sites

If you're comparing what the amps sound like compared to the JBL at the distance from your Helix in the picture, of course the IR is going to sound brighter. It's replicating a close-mic sound. Depending on the music and tones, a low pass that low in the highs might work fine, but it's probably not very accurate to what those amps sound like mic'ed through a PA. At the end of the day, though, it's all about getting the sound YOU want.

 

But that's actually the point, at least for me.  What I want coming through the PA should be relatively close to what's coming through my FRFR speaker.  Undoubtedly the close-mic sound of any of those amps would be different through the PA.  But the advantage we have with the FRFR speakers is we can have a more natural amp sound through the PA by making some adjustments in the signal chain to get closer to the sound we would hear in a room.  If it sounds great through the FRFR speakers, it should sound great through the PA (given you're working with a decent PA).

Share this post


Link to post
Share on other sites

I compared a cab block (separate to an amp block) with nothing else in the chain.

I did high and low cuts to taste in the cab block.

Then I'd also dropped a high/low-cut block right after the cab block. I duplicated the settings in both, and then I compared.

The cuts WHITHIN the cab block have FAR less effect on the results than the cuts following it.

That bears out my theory that the cab block cuts aren't as steep as a dedicated EQ.

I just realized; the other test I should've done to compare; put a duplicate high/low-cut block in between the amp and cab, just to see what that would do.

I suspect that would be more subtle than the other locations though, being as it would only impart its effects on the signal path up to and including the amp, master and all.

 

I found that, with the block FOLLOWING the cab, I had to drop it almost 3kHz further to get a similar result.

Are you saying that an EQ block hi cut after the cab has less effect than the same thing before the cab? Or is that sentence still comparing to hi cut within the cab block itself, which I strongly suspect is different processing altogether?

Share this post


Link to post
Share on other sites
post-10580-0-19878800-1487525857_thumb.jpg

 

I did another test this morning with my home-made FRFR (on the left). Its a Tremolux size cabinet with two Eminence Beta-10CX coaxial speakers with Eminence ASD:1001drivers. Its being driven by a Hafler Transnova P3000 stereo power amp, 150W/channel. The cabinet is ported in the rear. Here's what I found.
 
The EON610 and my FRFR behaved pretty much the same, they both needed the high cut. My FRFR is pretty directional, and certainly 2x10s driven by that Hafler is going to sound different than 1x10 with a horn driven by a 1000W Class D amplifier in that EON610. But they were certainly in the ballpark and needed similar high cuts.
 
I think the high cut on the Helix IR block is very gradual, and does not behave the same as an EQ block or Studio Pre with the same high cut. So the frequency numbers might not mean what we might normally expect for a typical 6-12dB (or more) low-pass filter.
 
The Helix IR block appears to be overly bright. My favorite IR with Helix is still the JOOSTALNICO-G12M-R121-U67.wav Joost created that's a mix of a G12M and a Weber Blue close mic'd with U67 and R121. Its a very warm sounding IR and doesn't require nearly as much high cut. See https://jimamsden.wordpress.com/2015/12/30/analyzing-jazzincs-srv-little-wing-patch/ for details. 
 
But the Celestion, 3sigmaaudio, Redwirez and other IRs don't sound overly bright when used with S-Gear in Logic, while JOOSTALNICO-G12M-R121-U67 is too dark. So I don't think its the IRs or necessarily the two different FRFRs. Rather it appears the Helix IR block is just a bright block and needs taming with the high-cut filter.
 
Further verification could be done by using recordings and headphones to eliminate the FRFRs. But at this point, I'm very happy with the tone I get out of the JOOSTALNICO-G12M-R121-U67 IR into the EON610 or my FRFR, and that was the point of the referencing. 
 
Frankly, Helix with JOOSTALNICO-G12M-R121-U67 into my FRFR sounds pretty good compared to that Showman. There's a transparency in the Showman that's not quite there with Helix, especially on the upper end. I suspect a small resonance bump in the IR high cut filter would fix that. That would tame the overly bright high end, but give a little bump right where the G12-65 starts to roll off quickly.
 
 

post-10580-0-19878800-1487525857_thumb.jpg

Share this post


Link to post
Share on other sites

So you're thinking the Helix IR loader is enhancing the high end... that's interesting. I'd think an IR is an IR no matter what's "holding" it.

Out of curiosity, what crossover is in those speakers in your homemade cab?

Share this post


Link to post
Share on other sites

The other possible conclusion is that the Helix's amps are brighter than your settings in S-Gear.

 

Is the Helix global EQ in play?

 

Would be cool to measure a completely neutral, flat IR, to know for sure if the IR block has some sort of high end emphasis. I'm not up to speed on making my own, but making one of nothing, with no processing at all, should result in something that  could be used for that test.

Share this post


Link to post
Share on other sites

Experiment: Load up the same impulse response in your digital audio workstation as you have in a Helix preset and A/B.

Share this post


Link to post
Share on other sites

(Sorry; quoting just does not work for me on this forum)

 

I was saying that yes, I suspect that putting the same high/low-cut block between amp and cab and after the cab would yield very different end results.

I'm yet to test that out; might do so this late late evening.

Share this post


Link to post
Share on other sites

But the Celestion, 3sigmaaudio, Redwirez and other IRs don't sound overly bright when used with S-Gear in Logic, while JOOSTALNICO-G12M-R121-U67 is too dark. So I don't think its the IRs or necessarily the two different FRFRs. Rather it appears the Helix IR block is just a bright block and needs taming with the high-cut filter.

 
Further verification could be done by using recordings and headphones to eliminate the FRFRs. But at this point, I'm very happy with the tone I get out of the JOOSTALNICO-G12M-R121-U67 IR into the EON610 or my FRFR, and that was the point of the referencing. 
 
Frankly, Helix with JOOSTALNICO-G12M-R121-U67 into my FRFR sounds pretty good compared to that Showman. There's a transparency in the Showman that's not quite there with Helix, especially on the upper end. I suspect a small resonance bump in the IR high cut filter would fix that. That would tame the overly bright high end, but give a little bump right where the G12-65 starts to roll off quickly.

 

 

I'd be a bit cautious about whether the IR processing in the Helix is different when comparing it to recorded IR's.  I know that recorded guitars in my Sonar sound a bit darker than my live IR's do through my DXR12.  I'm not saying it's not true, just that there are some differences beyond just the IR processing.

 

I agree with you that I'm less worried about that than I am just getting the right combination for my live performances.  I will reiterate that what I've experienced in terms of high cuts varies considerably based on the guitar I'm using, the amp model, and the particular IR.  For example I have a patch that I use with my Les Paul that incorporates a German Mahadeva amp with a Celestion Blue Alnico balanced IR that uses a Sennheiser MD421 mic that needs no high cut at all.

 

I guess bottom line it still comes down to your ears and simply knowing what to listen for when making these high cut decisions.

Share this post


Link to post
Share on other sites

 

 
Did some referencing today after repairing these two old Fender amps. The Tremolux on the Left is our bass player's old guitar amp. It needed an new bias filter cap. The Showman on the right is my amp from 1968. It needed a new output transformer, new tubes, new filter caps in the power supply and a new optocoupler for the tremolo circuit. The Temolux has the original Oxford 10K5's while the Showman has a pair of new Celestion G12-65's. That's an old Lexicon MPXG2 guitar effects processor on top of the Showman. Its still pretty interesting, but not used in this test. 
 
While testing the amps I thought I might not only compare the Tremolux and Showman, but also use them to reference my typical Helix patch, just to make sure I'm in the ballpark. I typically use that JBL EON610 as my vocal and guitar monitor so I used that to compare against the reference.
 
The Tremolux and Showman certainly sound different, I think mostly because of the different speakers and the different tremolo circuit. The Tremolux is the older bias tremolo while the Showman is the newer optical tremolo. The difference is that the optical tremolo cuts some of the gain in the Normal channel, even if the tremolo is off. That doesn't happen with the bias tremolo. Although they sound different, they are clearly the very similar sounding amps. That Tremolux has a little magic going.
 
Here's the surprise. I have the Celestion G12-65 IRs, and one I made myself of that exact cabinet with the Showman. I found using my IR and a Deluxe Vib amp model, I got an almost exact replica of that Showman amp coming out of Helix and that EON610. However, to do so, I had to turn the high cut on the IR way down to about 3.2 KHz. The Celestion G12-65 and all the other IRs I had were even brighter.
 
I was very surprised that all the IRs seemed to be a lot brighter through that EON610 than either reference amp. I found that I had to turn the high cut down to between 2 and 3 KHz to match the tone, and to achieve a warm pleasant tone from the Helix.
 
Have others noticed that the IRs tend to be very bright and need a lot of high cut in Helix? The 3sigmaaudio IRs were especially bright. They sound great with a lot of high cut, but I don't understand why this would be needed. I don't recall having this problem using the same IRs in S-Gear, but that's though different monitor speakers.
 
It could be the EON610. The response curve doesn't look like there's a lot of high boost, and I have the EQ turned off in the speaker. But maybe that horn is adding a lot of hype to the high end.
 
So the lesson is referencing real amps and then adjusting Helix IR low and hi cut to compensate for your FRFR can be a valuable and surprising learning experience.

 

 

Glad its not just me :)

 

Compared to the same IR in my Axe or Kemper, when loaded in the Helix, the require  more lo cut and more hi-cut.

 

When using my Axe or Kemper with my CLR for live rock shows - CLR is behind me angled up at me like a real cab - my  3 Sigma Zilla Bogner 2 x 12  IR's need  a  95 hz lo cut and a  5.5kz  hi cut

 

Same IR combo's in the Helix needed  110 hz lo and  3.8kz hi cut.

 

My gues is that the  lo cut and hi cut "slopes" on the Helix are tapped quite differently to the way they are tapered in the Axe and Kemper.

 

All the Celestion IR's I bought and tried made my 3 Sigma IR's sound " dull " ...... even a  110 / 3.8k  cut on the Celestions left them too bright for my use / ears.

 

And as for the OH's - I know some love them but  man are they  wooly / dull  compared to just about any other IR I have tried.

 

Ben

Share this post


Link to post
Share on other sites

I seem to recall a post by DI in which he pulled some code and found that the Cab block hi-cut was half as steep as the IR block hi-cut, or the other way around. Does anybody remember that? I can't remember when it was, but it would be right on point, except I can't remember whether it was the Cab block or the IR block that he said was steeper.

Share this post


Link to post
Share on other sites

Put a parametric after the cab on a Litigator patch I've been working on, left the EQ flat, copied the cab block low and hi cuts into the EQ, then took them off the cab. It was absolutely clear that the filters in the EQ are much steeper than on the cab. I needed to raise the hi cut significantly higher on the EQ than the cab to be anything like equivalent.

 

When l did that, it seemed there was a lot more freedom to raise the amp's treble and presence without hearing stuff I didn't like up top, possibly at a cost of seeming a bit artificial.

 

Net result was definitely different, I think in a good way, but I'm not sure. Will come back to it later w fresh ears.

  • Upvote 1

Share this post


Link to post
Share on other sites

It certainly would be helpful if Line 6 were to publish (more of) their specs.

Share this post


Link to post
Share on other sites

This might be worth a try, amsdenj... My band uses JBL 515XTs for mains. I use an EON 612 for my monitor wedge. I do have EON Connect working on my iPhone and have not had any issues with it connecting or communicating with the speaker. The other guitarist in my band uses Helix with an EON 610. Both the 610 and 612 sound very bright. I think it is because most other monitors we are familiar with, including the 515XT, have a natural roll-off in the high frequencies. Check JBL's website for response curves. EON 612 is one of the flattest I have seen. All that said, I set my tones (on Firehawk) to sound good through the 515's since I know that is what the audience will hear. Here's the kicker. I use a high cut at 4.5 KHz on Firehawk. Why 4.5 you ask? Because that is the fixed high shelf frequency of the EON's custom EQ. I run the high cut on Firehawk at -3.5 dB on most of my patches to get them sounding good on the 515. Even so, I still need another -1.5 dB to get rid of the fizzys when monitoring on the EON. I do this using the EON's custom EQ. I don't have a Helix (yet), but I posted here because I think this effect would translate to any modeling source into the EON. If you setup your Helix's global EQ to sound good on the 612, you may be cutting too much in the high frequencies for other monitors.

Share this post


Link to post
Share on other sites

I did some more quick testing tonight just to verify what I was seeing in the live setup with the EON610 - that the Helix IR is pretty bright and needs a fair amount of high cut. I wanted to verify the observation without having any FRFR speakers involved at all.

 

I created a patch in Helix that had just a Deluxe amp model and IR block. Then I created threes tracks in Logic Pro X and recorded the same quick, short guitar part three times:

 

Track 1: Deluxe block + Helix IR block with Celestion G12-65 212 C R-121 Balanced IR, with high and low cut off

 

Track 2: Deluxe block only, but with a convolution plugin in Logic with the same G12-65 212 C R-121 Balanced IR, no high or low cut.

 

Track 3: Same as Track 1: Deluxe block + Helix IR block with Celestion G12-65 212 C R-121 Balanced IR, but with high-cut set to 3kHz

 

What I found was what I expected: Track 1 was quite a bit brighter, Track 2 sounded like I would have expected the IR to sound, and Track 3 was pretty close to Track 2. That is, the Helix IR block appears to be overly bright and high-cut somewhere between 2 to 4 kHz is required to reproduce the intended IR output.

 

Now I don't know why this is. I suspect its because convolution algorithms are computationally expensive and can introduce latency. Line6 may have a 'optimized' algorithm that takes some shortcuts to save compute cycles, and this could result in the block not completely following the IR. Since guitar speaker IRs generally have a lot of high-end rolloff, perhaps an optimized algorithm would not completely follow the IR and would let some of the excessive highs or dry signal come through.

 

Note that I also tried a number of different convolution plugins in Logic including SpaceDesigner, LAConvolver, NadIR Convolver and mixIR2. These all sounded a little different too. So Its not surprising that different convolution algorithm implementations might sound a bit different.

 

This was a quick, simple, informal test; just to verify initial observations. More careful analysis using pink noise or a sine sweep and a spectrum analyzer would likely be more accurate than my old ears. But at this point, I don't need to do any more analysis for my purposes - which was to simply verify that I am getting the most out of the IRs I purchased, and the tone I'm getting from Helix references well with a real amp.

 

It would be nice if Line6 could provide some additional analysis of what some of us are informally observing, and some recommendations on how to best use the IR block and its high-cut control.

  • Upvote 1

Share this post


Link to post
Share on other sites

Thanks amsdenj, that might go a long way to explaining why some of the distorted amps can sound overly 'fizzy' without the high cut filter in place. And the problem with the high cut filter is that it doesn't roll off steeply enough to kill the high end properly without starting it so low that it starts to kill the cut and sparkle of the amp.

 

Convolution shouldn't be something that has 'flavours'. Obviously, if an IR is too long then it needs to be truncated which will change its tone and characteristics. But, absent truncation, the only differences should be incredibly subtle rounding errors during the FIR or FFT/IFFT processing. Even if the overlap add/save methods are used, the FFT based convolution should be mathematically equivalent to just using a really long direct convolution with an FIR kernel. There are no window functions here, so no subjective trade offs or choices that may impact tone. Internally accumulated rounding errors in the FFT/IFFT may cause an increase in the noise floor if the processor is relying on 32 bit floats with a 24 bit mantissa. But, if they are using 32bit integer arithmetic with a block floating point exponent, then the noise will be insignificant.

If the output of the IR convolution isn't completely serial (ie. mix level isn't really 100% IR and a little bit of raw signal is leaking through) then this would impact the expected frequency response of the IR in a way that would make any attenuation (at any frequency) seem far less dramatic than it should be. ie. at a 50% mix, an IR that cut a given frequency band by 90dB would appear to only cut it by 6dB as the original signal is mixed back in at 6dB down after being cut out.

 

When using cabinet IRs, make sure the blend is set to 100% to avoid this problem. If there is still a problem, then maybe there is a bug that is causing some raw signal to bleed 'around' the IR.

Share this post


Link to post
Share on other sites

When using cabinet IRs, make sure the blend is set to 100% to avoid this problem. If there is still a problem, then maybe there is a bug that is causing some raw signal to bleed 'around' the IR.

I'm ignorant about convolution maths, but on a gut level, that's not what I hear, I think. Just seems like the roll-offs aren't very steep, not that there's unfiltered signal getting through regardless of filtering.

 

So we have two questions here:

- Is IR processing in Helix somehow different from other IR loaders?

- Are the Helix cab and IR hi cut filters able to reproduce the behaviors of nicely mic'd real-world cabs?

 

One interesting bit of research I don't have time to do would be to examine the published response curves of various guitar speakers (ideally in some typical cabs, though that info may not be as available, and I'd bet the cab doesn't affect highs as much as mids and lows), and see how steep the hi end roll-offs actually are that we're trying to mimic.

 

Really, since we're talking about the whole recording chain, the mic and mic pre are also in play, and since we're talking about "standard" recording industry practice (yes I know there is no such thing), typical mix processing is also involved. Each of those elements has its own inherent and/or commonly used hi frequency limits, what we're used to hearing is the net of all that.

 

If I'm not mistaken, the simplest electronic filtering (a cap to ground) is inherently 6db per octave, not very steep, don't know if that's true in the digital domain.

Share this post


Link to post
Share on other sites

A typical 12" guitar speaker rolls off at around 5 to 6kHz at about 20dB per octave. It has response from about 7kHz to 20kHz that bottoms out about 20 to 30dB below nominal with some unevenness and a very directional response pattern. Roll off in the low frequency side is more gentle at about 10dB per octave and starts to drop off around 70Hz to 100Hz being about 30dB down at 30 to 40Hz.

 

So an IR with equivalent frequency response should have some decent roll off just like a real speaker, and should not need extra filtering if the IR is doing its work properly. If the IR output is not at 100% blend (ie. some raw signal is being mixed in), then it will seem smoother and the high and low frequencies won't be attenuated anywhere near as much as would happen with the IR by itself. This is irrespective of any high/low cut filters.

 

It is possible to apply a filter to the IR to create the desired roll off if the natural speaker capture wasn't enough. However, what has been observed is that the same IR used in a DAW IR convolver is a lot less bright/harsh than the same IR processed in the Helix. Since it is the same amp and signal path, the only things left are either something wrong with the IR processing in the Helix (or the DAW), or there is some raw signal mixed in parallel with the filtered output in the Helix. It should not be necessary to add extra low/hi cut filtering to get the same frequency response in the Helix as happens in a DAW. The convolution algorithms should either be the same or mathematically equivalent. Anything else is either user error (possibly because of counterintuitive/hidden default behaviour), or a bug in the Helix.

 

There's more than enough frequency resolution in a 1024 or 2048 point IR to map the high frequency response of a guitar speaker to the point where a signal convolved with the same IR should be indistinguishable between the Helix and a DAW. If there's a difference, the source of the problem should be determined. Bandaid solutions like adding low/hi cut filters should not be necessary to get an IR response to match between the Helix and a DAW.

Share this post


Link to post
Share on other sites

What I found was what I expected: Track 1 was quite a bit brighter, Track 2 sounded like I would have expected the IR to sound, and Track 3 was pretty close to Track 2. That is, the Helix IR block appears to be overly bright and high-cut somewhere between 2 to 4 kHz is required to reproduce the intended IR output.

Something's definitely not right there. Do you mind posting pics of the signal routing and clips of the recordings?

Share this post


Link to post
Share on other sites

I would have to do this test over again, give it more time and care, before posting anything. I did this very quickly last night and could have made some mistake, but I don't think so. The IR block in Helix and the DAW are both set at 100%, so there is no user set bleed of the dry signal.

 

Perhaps some other people can try this experiment and see if its something I did wrong, something unique to my Helix/DAW, or something wrong with my ears (or I got biased by what I heard referencing with real amps and the EON610).

 

Independent verification of a quick observation might be more valuable than me doing it again.

Share this post


Link to post
Share on other sites

A typical 12" guitar speaker rolls off at around 5 to 6kHz at about 20dB per octave. It has response from about 7kHz to 20kHz that bottoms out about 20 to 30dB below nominal with some unevenness and a very directional response pattern. Roll off in the low frequency side is more gentle at about 10dB per octave and starts to drop off around 70Hz to 100Hz being about 30dB down at 30 to 40Hz.

 

So an IR with equivalent frequency response should have some decent roll off just like a real speaker, and should not need extra filtering if the IR is doing its work properly. If the IR output is not at 100% blend (ie. some raw signal is being mixed in), then it will seem smoother and the high and low frequencies won't be attenuated anywhere near as much as would happen with the IR by itself. This is irrespective of any high/low cut filters.

 

It is possible to apply a filter to the IR to create the desired roll off if the natural speaker capture wasn't enough. However, what has been observed is that the same IR used in a DAW IR convolver is a lot less bright/harsh than the same IR processed in the Helix. Since it is the same amp and signal path, the only things left are either something wrong with the IR processing in the Helix (or the DAW), or there is some raw signal mixed in parallel with the filtered output in the Helix. It should not be necessary to add extra low/hi cut filtering to get the same frequency response in the Helix as happens in a DAW. The convolution algorithms should either be the same or mathematically equivalent. Anything else is either user error (possibly because of counterintuitive/hidden default behaviour), or a bug in the Helix.

 

There's more than enough frequency resolution in a 1024 or 2048 point IR to map the high frequency response of a guitar speaker to the point where a signal convolved with the same IR should be indistinguishable between the Helix and a DAW. If there's a difference, the source of the problem should be determined. Bandaid solutions like adding low/hi cut filters should not be necessary to get an IR response to match between the Helix and a DAW.

That typical roll-off info is great, thank you @jnysen! Where did you get that info? Nice work.

 

Leads to some questions and conclusions:

  • However the built-in cabs are implemented, I think it's clear that the hi cut filters aren't anywhere near that steep, on cabs or IRs. Ideally they would be, if they're intended to adjust cab tone while "keeping it real". That could be responsible for the high frequency artifacts that bother some people. Even the hi cut in the parametric EQ, which seems considerably steeper than what's on cabs and IRs, doesn't seem like 20db/octave.
  • I wonder if the "optimization" that Line 6 has said their cabs have over standard IRs results in less steep rolloff.
  • The point you make about "should not need extra filtering if the IR is doing its work properly" applies both to built in cabs and IRs. I've always wondered why any sort of filtering was required to make a cab sound like what the cab is a model of, and the same is true of IRs.
  • If those slopes are pretty consistent across various cabs and speakers, what's responsible for the differences between them? Is it mostly in the narrow-band "uneveness" you referred to?
  • Many physical systems that have some sort of roll-off have some degree of a resonant peak before that, and I'm pretty sure that's true of speakers. It'd be great if the Helix filters let you control that.

Share this post


Link to post
Share on other sites

The IR capture process affects the amount of high-end you'll hear in the resultant cab IR.

 

With the right front end gear (and patience/experience), you can capture an accurate/detailed cab frequency response that needs no high-end roll-off.

 

 

Note also that some convolution plugins might limit frequency response to save on computation.

ie: Rarely do you want to hear reverb go up to 20k.   ;)

Share this post


Link to post
Share on other sites

I mostly use an Ownhammer Mesa 412 IR and I don't run any cuts on it. However, the stock cabs I usually have to cut pretty aggressively if I have any distortion going through them. I have also found that the numbers on the amp controls on the Helix don't really seem to correspond exactly with the real amps, or at least there is not a terribly great way to line them up. So it's extremely difficult to just create a patch based on numbers alone.

Share this post


Link to post
Share on other sites

Hi all

 

Just want to chime in here again with some further real-world gig experiece  re: the  hi / lo cut  "issue".

 

Long winded  Background .... sorry :(

 

Had a show last night and got to use the other guitarists Helix as he was double-booked for another show elsewhere - was me and another of the "regulars"  -  all good.

 

We did  3 x 45  "classic rock"  ..... you know, all the ususal suspects   :).

 

Prior to the show  - I loaded up the Helix presets I made when I  had  "his"  Helix   :) ..... no problems ... all good.

 

For all my live work I run a CLR Active wedge behind me - and I use either my Axe XL  or  my Kemper - the Axe and Kemper use the same IR combo - 2 x 12  3 Sigma Bogner Zilla  3B  and 5A  in a 50/50 split  summed to 0 db with a  95hz lo cut and a 4.2khz high cut ..... strongs cuts I know but it gives a  "really real"  guitar cab vibe   :)

 

By way of backgrounf - in the Axe I use a  6 db slope  with the  95 / 4.2 cuts ...... with the Kemper I  pre-mised the 2 cabs externaly and matched the hi / lo cuts by ear and recording  [ Kemper doest call their Cab hi / lo cuts  ... its just  + / -  db adjustments.

 

Anyway .... tonally  [ not gain-wise ]  the Axe and Kemper  presets and profiles all sit in the same nice "range"

 

With the Helix last night, I ran the same to cabs,  50 / 50 split summed to 0 db set to  95 / 4.2k cut ... same as the same cabs in the Axe and Kemper ...... note guitar  [ Tele ]  is my regular across all rigs.

 

Helix sounds were  amazing  live ..... genuinly great ... that was as I had expected .... the problem was that I had these  lo / hi frequencies that had me dumbfounded ..... was like I had a low / high boost  in the presets ... whcih I knew I didnt.

 

30 min break between sets !!!!

 

End of set 1,  took a 10 min break - grabbed my ATM headphones and wen to my rig.

 

Plugged in .... even in my headphones it was these - I checked all my presets - all good.

 

As a last effort  I went to my IR slots .... damn they were set right too ..... 95 / 4.2k.

 

About to go back on in 10m ..... why not I figured ... I randlomlny re-set all the Cab Lo / hi  cuts  to  120hz and  3k .... worst case scenaerio I thought .... Id rather less lo / hi end than more if they were my only 2 choices.

 

Bingo ......... the 120 / 3k cut sound  pretty much just like my Axe / KPA  tonal ranges

 

Got home ... did some further "testing" ....... this really was  "bugging me".

 

Result

 

Same IR's..... same settings .....  95 / 4.2  cut in Axe and KPA = what my ears know and are used to .....  Helix ended up needing on   ~110 / ~3.2k  cuts  to sound  "tonally"  the same .....  however ***

 

Conclusion

 

From my so-far limited expereice and so-far basic test ... my ears are telling my that  the  Helix Cab  Lo / Hi  Cut   " slopes'  must be  realtively  quite flat .... like I said ... in the Axe I use a 6db per octave slope  [ it also has a 12db per octave slope from memory ]   and this  6db per octave slope is very easily duplicated in my Kemper.

 

I can only assume that that the  lo / hi cut  slopes in the Helix are of a totally differnet "shape" / "structure".

 

***  -  in a practial tweaking sense, the real issue with flatter-slopes  is  mor noticeable in the hi end  ie:  even a relativley small change from  say  3.2k hi cut to  3.4k hi cut makes a BIG  difference to high end content re-introduced .... similar, but not as audibly  immediate in the lo end too

 

Anyway .... 2nd a 3rd sets sounded awesome - as good a sound as Ive gotten from any modeller ....... but would be really good if this  "issue"  could be addressed  "up the line" ..... ie:-  selectable  [for example]   6db per octave and  12db per octave slopes for the lo / hi cab cuts would be ideal

 

Thanks to all.

 

Ben

  • Upvote 1

Share this post


Link to post
Share on other sites

This is interesting. I'm guessing a too subtle slope would explain why when I try to tame the krinkle to where I like it the the overall tone just gets too low middy and kinda dead. ?

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...