What is a proper input level? And how to choose the right impedance?

I always wondered what a good input level would be to begin with. Trying out different presets by other users I noticed they sometimes sound way off. When adjusting the input, most of them turned out good. So I guess many users are dialing in very different input levels.
Is it better to reach a level that is close to clipping (close to -0dB and thus turning red at peak volumes of your instrument like on a typical interface)? Or should one rather stay at lower volumes so the preset and the blocks within have more room to play?

And obviously the impedance does matter too. But how to decide on that? And please don’t answer with the obvious “use your ears”-phrase. :wink:


2 Likes

IMHO the level should be as high as you can get away with.
The goal line is the small break in the bar you can se in your first screen shot.
Anything above that will cause clipping.
I always hit the lowest note harder than I would playing regularly and adjust the input to be just before or slightly clipping, so I have some headroom.

Worked like a charm on the Helix and on the QC.

As for impedance I played around with it and didn’t notice a drastic difference, so I just left it at the default 1M Ohm.

1 Like

In the analog/tape world, you always tried to keep the signal as hot as possible to get the best signal-to-noise ratio. And a little analog clipping was actually a bonus. The digital domain is the opposite. With 24-bit A2D and D2A, you don’t need to worry too much about the noise floor and digital converters don’t introduce noise as tape did. However, digital clipping is something that is to be avoided at all costs. There is almost no such thing as good digital clipping (bit-crusher might be the exception if you’re into that sound). So being conservative on the input level isn’t really an issue. Better to be a little lower and avoid any digital clipping than to focus on driving as hard as you can.

QC inputs set at 0dB will pretty much take a Humbucker pickup flat out without any issue. So unless you have unusually weak or strong active pickups, I’d set input to 0dB and not worry too much about it.

11 Likes

So two answers with two different views so far. Exactly what I hoped for :crazy_face:
I noticed that the signal coming out of the QC with no effects / cabsim / etc. is about 5 dB weaker then when going right into an amp / interface.
So I have the input set at about +5dB to get the same signal level as I would not using the QC. Then preamps, compressor and drives (and amp+cabsims if used without a real amp and cab) will bring it up about another 10-15dB.
To me this sounds very much like my analog pedals sounded. However when creating a capture, I set the input as hot as possible to get the best results. And that is really weird to me.

1 Like

Another perspective:
Initially I boosted the input, just like you’d do with an interface. But then all the amps behaved very odd. E.g. a DLX or Super Reverb model still distorted at Vol around 2, with vintage single coil pickups. And overall, every effect and/or amp model clipped way too early, it was impossible to set up my signal chain like I would with analog gear.
Then, I set the input to ± 0 dB, and my life changed :slight_smile: Now all the models in the QC perform and feel pretty much like the analog versions they’re modeled after.
So I suggest, pick an amp model that you know very well from real life experience, and adjust IO such that it does what you expect it to.

4 Likes

I just got my QC, and at first I used the input meter and just made sure I wasn’t hitting the red, cuz I know that’s not cool in the digital realm (as opposed to analog)….but in getting to know the unit just by diving into presets, the high gain stuff was SO fun to play, but I was afraid I was veering toward “insane mode” territory. Cleans were terrific, but low/mid gain presets were kinda ugly. I just decided to try lowering the gain to +4db and I’m already loving this thing even more than I thought possible.

I started my first preset without worrying about that.

It sounds very good. It’s a Friedman with gain at 4. Then I add a drive block (OCD). Fine.

A few days ago I started checking the Input Level.

I find that when I get close to 0 (input level around 8-9), well, the amp model sounds good, but when I add the OCD, there is a lot of compression smashing the sound and I feel it sounds lower.

So I’m going back to default.

I also don’t see any difference when moving the impedance knob.

Just my first impressions.

this is a topic you can’t really answer without the ‘use your ears’ clause. Your instrument, pickups (mic), setup, cable etc- can make so much difference there just can’t be one recommended level in general. The one thing I’ve discovered from experimenting w/ NDSP products is not to be afraid to turn things DOWN; as mentioned in other posts lowering the input levels can greatly improve your clean tones

1 Like

I understand and respect everyone’s answers, but I feel that that there is a strong case to be made for having a reference standard. The main reason being consistency. If every user is setting their input gain to a relatively similar setting (meaning the signal strength, not the gain value) then patches would translate better from user to user.

I would also suggest that Neural certainly used a standard of some kind when creating the Quad Cortex, so it would be helpful to know what the intended input gain is for the amp models and effects.

2 Likes

Hard strums peak at -12db. 1M impedance unless using certain fuzz pedals first in chain.

0dB input gain would be the reference standard. That’s unity gain, not boosting or cutting your guitar output. You would typically only change this if you had a problem instrument with unusually low or hot pickups to avoid possible input digital clipping.

I never touch the input gain as I want patches to respond differently to different guitars.

3 Likes

You would maximize analog to digital input conversion until clipping, ALWAYS (with a margin for reasons too complicated for a forum post, I use -2db for the highest peak).
If you do not maximize it you will lower the snr, because the noisefloor digitalized is still a noisefloor and if you sample with a max amplitude of say -24db instead of my -2db you are actually losing that 22db of snr (you should raise the final volume to reach the sound pressure desired so also noisefloor will raise). The guitars do not produce such a snr (mine is well under 90db) so we can afford to lose some snr, for synths is really different!
jamsden is right saying that modeled amps will respond with fidelity to a certain guitar only with no added gain…but if your goal is to record with max snr you should maximize the qc input ADC and THAN lower it digitally by the same amount to have the matching response (usually the gain inside amps block does exactly this, but can be necessary to further lower it with a gain block).
Please note that adding gain over certain level can raise the analog noisefloor that will be digitalized, the best compromise is different with each case. Also QC gives its preferred level (strangely low), so in the particular case it cuold be better to just use that advice from the developers. Probably all the simulations expect that level to sound “right” at defaults.

Maximizing input gain is the recommended practice for analog systems, especially when recording to tape. The reason is to achieve the best signal-to-noise ratio when using a device like a tape machine that tends to generate a lot of noise. Also analog devices can sound better with a little saturation, so there’s not much danger of sending a signal that’s a little too hot.

But this is not true for digital systems for two reasons. First digital systems don’t generate much noise on their own. So with 24 bits, you don’t need to worry as much about noise floor. Second, there is no such thing as good digital clipping. When the A2D or D2A converters run out of bits, they clip badly and not in a musical way. So its really important to not clip the A2D or D2A inputs and outputs. This is why QC has a limiter on the output, to avoid any chance of digital clipping.

So the convention with digital systems is to worry less about noise floor and more about avoiding digital clipping. Recommended practice is to have 12 to 18 dB of headroom when recording.

3 Likes

Hi jamsden (are you the ambient guitar guy on yt btw?)
I agree with you, all you said is good advice.
ADC/DAC is my daily bread an butter so I just use way less safe zone because I know how to do it. My QC is just at +10db using 8db of average headroom indicated by the vumeters onboard, 2db only when going crazy on strings with the highest pickup selection (at the raw digital file produced). Is a little too hot for most of simulation but I am what I am, I prefer a gain reducer to a worse snr on my DI, I am not pretending that is how to use the QC.
I think the argument is way too complex if we start talking scientific and out of the purpose of this thread but I enjoy it, if you want we can talk about that in a private way.
Also I would like to play some ambient bass on the side of your guitar…if you are that guy… we can share preset to be in the same ambient, should be wonderful!

1 Like

I am getting a strong feeling that these input level things and challenges with clean sounds are actually HW related and to be more precise, difference in HW. See my challenge in the topic linked below:

I have been in contact to Neural but no solution yet. I have expressed a worry that not all HW are the same.

In my case I need to use ~ -12dB gain to get clean sound with a clean sound factory Preset, with otherwise Factory Settings (other than the input gain). This happens with all of my humbucker guitars. And I am not the only one, even on this particular board having such challenges with clean tones. With distorted sounds this is more difficult to notice… naturally. So with factory settings, with factory presets I get clipping even without having the meter going to red. To get the clipping gone, I need to go so low with the input gain that I have ~20dB of headroom, looking at the meter. I am getting strong feeling that there are differences with the input stage of QC’s. Just thinking about the serious component shortages they had at one point of the deliveries. What if other persons +10dB is the same as other persons -2dB. To me it starts to seem possible. That would explain a lot of these challenges and concerns.

Unfortunately, every piece of audio hardware (and software) is different. Most manufacturers leave a small buffer for occasional transients that stray into the red. My ol’ Fractal AX8 manual recommends setting the input gain that way. Some guidance from the folks at Neural would be helpful. Is everything in the red actually into digital clipping?

Overall, I found the QC manual to be an easy read but very “nuts and bolts”. Perhaps they’ll start releasing supplemental documentation later to fill in the info gaps and help us make the most of this amazing little tone monster.

I dont think this is a what if, but rather an absolute. It explains the majority of issues people have regarding “good sound”. I think the mistake (myself included) is the thinking that the cortex is the same as a real amp, despite sounding like a real amp.

Yeah. So I saw this last week and it explains digital modelers’ input level vs amp input and how some instructions can be misleading to users:

It’s pretty long and very thorough and focuses on the Helix & Tonex but the same principle applies to pretty much any modeler. I think this could be useful to many users!

1 Like

I sincerely doubt that there would be such a huge difference between units. Firstly, I would not assume that the factory presets are dialed in to be perfectly squeaky clean, especially with humbuckers. Things usually sound better with a little breakup to give a bit of fatness and compression. Individual differences in playing style and instruments will also factor in.

What happens if you turn down the gain on the amp block? Does it clean up? Do not worry about the knob position or the number on the dial, just use your ears. You may need to bring up the output gain on the block to get back to the same volume level. If they simply do not clean up even if you adjust the gain then there must be a problem elsewhere. Perhaps an actual defective unit.

At 30:02 he says… “This is irrelevant. This does not matter.” I found that to be hilarious.

Good video. Ending has some tasty jams too. Thanks for the share.

Regarding capturing i’ve found that zero gain (0db) for level 1&2 sounds the best to my ears. Straight up ignore the bars. Lol. Started capturing my Mesa Prodigy Amp only (through the DI) using this. Sooo good. But not sure why.

1 Like