Setting the input gain on audio interfaces

As some of you know I only play using amp simulators (stealthy bedroom player and so on!)

I have tried several interfaces and I always seem to have the same problem: I can’t raise the input gain to the suggested level (to hit the “green band” of my amp simulator input level) without clipping the signal.

Would a DI box help? Can a cheap one do the job?

EDIT: here’s a visual summary of the issue (copied from my post below):
EDIT2: The interface in this example is a Scarlett 2i4 2nd gen., the AmpSim is Overloud THU


Tommo, first the guitar signal goes into the interface. Are you able to get a good signal there? Meaning, does it light up the green light on your INTERFACE (not in the plug-in), with maybe an occasional orange light at the loudest notes?

1 Like

Hi tommo,
afaik many interfaces have an instrument input you need to use if you plug the guitar directly into the interface. This is called Hi-Z-Mode or Hi-Z Input. Sometimes you need to activate that mode in the driver. This is the case with the RME Babyface I use for example.
There was a great explanation in a recent Neural DSP video for that kind of situation, here is the link to it:

1 Like

Thanks both!

Yep, this happens, but it’s still far below the “green” threshold of the plugin (Overloud THU in my case). This is what should look like, while I am stuck in the “deep yellow” :smiley: unless I allow for some serious clipping (red light) at the interface:


@Bastian93 thanks for the video, will check it out! I do use the “instrument line in” setting on the interface, but I never thought I might have to control the impedance via drivers - will see if I can do that tonight!

PS: my main interface is a Focusrite 2i4 2nd gen, if that helps!

I have a Focusrite myself (2i2). First indeed check if you have the lastest drivers. Then check (in the Focusrite manual) if the impendance is correct for your guitar at the input. I suspect it is. Then adjust so you get in the green with the Focusrite.

If this all works, then it’s the plug in that is causing problems.

You can download a trial version of amplitube or something similar to check if the input signal is ok with the other simulators.

Check if your current sim has an input gain button (they always do).

Good luck

1 Like

I’m not sure that I understand the problem. Your guitar is an analog device that has one ground and one signal that varies. A DI is an analog device that will give you a differential signal (three wires, where one is ground and two are differential, it does this by means of a transformer). As you know (being a physicist), differential encoding is really smart because it is immune to common-mode noise, but guitars are badly designed regarding electronics. In general you’d use a DI for pro audio, as they want to keep noise down and like differential.

The next step is that the signal goes to an ADC (analog-to-digital converter). Over here you might be losing a few bits of signal if your output isn’t hot enough, but it probably won’t matter if it’s calibrated appropriately (so the “virtual amp” knows if it is being “overdriven”).

What model of equipment are you using? It almost sounds like you want some boost, and there are pedals that do it, as well as in-guitar devices like

I have the opposite problem than you, my guitars are so hot (they all have opamps inside via EMG active pickups) that they saturate the inputs on my wireless, my Axe-FX, whatever.

Thanks for the reply @kgk

Actually it seems to me that we have a similar problem, even though my pickups are all passive.

In a nutshell, I would like to use the interface gain to provide a strong enough input to the amp simulator - but at the moment I can’t do it without clipping.

So I am pretty much stuck with a weak signal going into the ampsim. I guess I could compensate for that within the software, for example by putting a clean boost plugin (ideally flat in frequency response?) before the amp simulator plugin.

But I have this semi-irrational feeling that optimising the DI signal (before it goes into the ampsim) via hardware would be better :thinking:

Is the interface analog? Is the output of the interface analog or digital? Where does ADC take place? Which part of the signal chain is clipping? (One can clip in both the analog and digital domain.)

This happens inside the Focusrite 2i4 (not sure of the technical details)

Here I am not sure because all I get at the front of the interface is a generic “clip” warning.

The situation looks like this:

Ah, I see: You send your guitar to the Scarlett device, and its input turns red. So, what does “turn red” mean? It looks like there is a pot that is used for analog gain, and then it likely goes to an ADC. So the ADC is saturating (using maximum or minimum integer that it can encode), and then the Scarlett’s firmware probably notices that it has “too many” such saturated samples in a row, and as a consequence it lights up the ring to warn you… that part seems reasonable. From there on, you’re all digital, and that’s what you feed to your modeling software, likely by a serial bus (USB).

So your “amp” software runs on a laptop and THAT software has the error message that you’re not playing loud enough, unless the Scarlett is reporting clipping.

Who makes the software? My guess is that they’re looking at the average power and don’t really care about clipping, because there is a huge difference in dynamic range between smashing all six strings in chords like a folk singer vs. tapping a single string, so they want to be sure that the input is reasonably normalized. They probably expect clipping at the front end and don’t care.

Does that make any sense? (In other words, don’t worry as long as you don’t see the red light when you’re paying your normal music.)

1 Like

Thank you, it does :slight_smile:

Interestingly, what seems to clip the most is pinch harmonics on the low strings, it gets me in the red zone much more consistently than madly strumming a 6-string chord!

I also forgot to mention that my ampsim of choice, at the moment, is Overloud THU. It’s just too convenient to have effects, amps and IR loaders within the same plugin.

(Although I just came out of a pathological GAS period in which I bought several others, including all the Mercuriall and Audio Assault stuff).

PS: and I totally agree with you on your previous posts about guitar wiring. It would be amazing to have a stereo output capturing two pickups at once - among other things this would generate so many mixing/re-amping possibilities!

I’m not sure the instrument level inputs on the focusrite interfaces are particularly well thought out or implemented - if you can take care of your first stage of clipping before you get to the interface (i.e., stick a pedal infront) I think you might get better results. Otherwise a fet-based di like a countryman 85 will also be good.

I think problems arise because the dynamic swing between your most powerful output (usually a muted E powerchord but maybe a pinch harmonic beats that), your sort of middle volume regular playing and your quietest playing is so great that there’s just no way to set up the interface and the amp sim that you get good results across the whole range. Better to compress your signal first in some way whether that’s a clean boost, light drive, compressor whatever.

Even even better in my view is something like the AMT legends series preamp pedals - run the guitar into that, let your main tone generation be done in the analog world and then run your cabinet and room IRs in the DAW along with whatever post effects you fancy.

If you do want to stay purely in the box you could try ensuring that even the highest level out of the guitar doesn’t clip the interface then use a gain plugin (like Utility in Ableton) to make up the gain before reaching the amp sim.

1 Like

TLDR;avoid A-D clipping and D-A clipping at the master. Look if there is a software mixer between analog input and the amp sim.

Is there some sort of software mixer “in-the-box” before the guitar sim software, that is lowering the gain your amp sim sees?

The reason I ask, is because that is how it is with my RME Babyface pro. So my setup is, never clip the analog input, if needed I can boost the input in the software mixer or in my plugin, S-gear. Inside of s-gear I can turn up the input gain all the way up, if I want to, and drive the “front end” of the amp as hard as needed. I do take care not to clip the output, if I am using S-gear as a stand alone app.

Inside a DAW clipping it is sort of the same, avoid analog clipping and clipping the master output, before D-A.

Bonus question : does it sound good if you are not clipping the analog in and simply raise the gain on your “amp” inside the sim? Or is it horrible? No reason to fight a problem just because the manual said so, if it sounds good.

1 Like

Your number one goal is just to get the signal into your computer without clipping. Do some loud strumming, or loud rhythm playing that represents about as loud as you might realistically play. Set your input gain so that this loud stuff doesn’t clip and those loud peaks end up somewhere between -20 and -10 in your software. You’re done.

Longer answer: There is no need to squeeze super high up against zero on modern gear with decent or better preamps and converters. In fact, pro-level digital gear was originally designed so that approximately -18 (give or take) in your software meters was the equivalent of zero on an analog mixing desk. Meaning, your interface would be designed so that when you set the gain in the sweet spot of what the preamps and converters can handle, the digital signal that the device outputs to your software ends up around -18. The idea was that everything above that, between -18 and 0, is basically insurance that equates to the “headroom” you have in analog gear where you can come in a little hot and not clip. The exact numbers will be slightly different depending on the gear you use, but I try to end up somewhere between -20 and -10 just to keep the analog gear running in the range where it was designed to sound best.

Once you’re in software, you now have a clean non-clipped signal. If your amp sim software is expecting something louder, then you can just chuck a compressor on it before the amp sim, with gentle settings to take down some of the highest peaks, so that you can turn the makeup gain (i.e. compressor output) higher and still not clip over 0. However a good amp sim shouldn’t really require this. Amps are (among other things), compressors, and the amp model you load up should take care of all of that without requiring you to slam its input with a signal that is kissing zero on your hardest strums. If it does, I’d do the compressor method and listen to the output of your amp sim to make sure it sounds good.


Yeah, Troy has the right of it - in the digital world, your #1 priority is not clipping on input. You have 100% transparent noise-free gain staging available to you “in the box” once you’ve turned your analog signal into digital, so bring that in as cleanly as you possibly can. If necessary if the DAW is still showing you have ample headroom after the .wav file is in a track in your DAW, you can adjust the volume of the raw track there either through normalizing or some sort of variable make-up gain - Reaper lets you do this by right clicking and selecting some sort of properties menu which has a number of gain options. The Overloud plugin might also have some sort of trim control you can use to boost the signal as it’s coming into the plugin.


Thanks everyone, this has been instructive - and saved me money :slight_smile:


On a real amp actual voltages matter (to enter nonlinear operating regions), but on digital systems there is often little bearing: For example, my wireless pads my guitar by -12dB (likely because I have active pickups), then cranks it by +24dB, and then I have to drop it down at the input of my Axe FX to “20%,” etc., so how does the software “know” if I’m playing loudly [non-linear], or not [linear]? I believe that what they do is normalize compared to a reference that is specific to one’s guitar (somehow set by heuristics with the “yellow light”) and then reason relative to that (so they’ll likely multiply by the reciprocal of the reference). I wish they were more transparent about the process, but I think that they don’t want to scare customers.

I’m not even concerned with clipping as long as it’s rare: I love distortion and likely won’t tell the difference anyway. I think the real reason not to clip is for purposes of re-amping, e.g., one decides to turn of the (fake) “Mesa Boogie Dual Rectifier” and go “acoustic.”

1 Like