Why do amp models have to keep the unwanted stuff?

Discussion in 'Digital & Modeling Gear' started by TonePilot, Mar 12, 2018.


  1. vtgearhead

    vtgearhead Member

    Messages:
    570
    Joined:
    Aug 24, 2016
    Many years ago, noted audio designer Bob Carver (Phase Linear, Carver Corp. and others) ran a set of double-blind listening tests to see if "golden eared" audiophiles could be fooled during A/B comparison between tube and transistor amplifiers. Long story short, it turned out that part of the magic formula for emulating a "tube" sound was injecting very small amounts of hum and noise into the program material.
     
    TonePilot likes this.
  2. TonePilot

    TonePilot Supporting Member

    Messages:
    446
    Joined:
    Feb 28, 2018
    Location:
    Canada
    The models are noisy even without my guitar
    I’m really curious how you know about the electrical interference problems on the ISS. Something in outer space with no magnetic field to protect it. I’m both impressed and highly skeptical.
     
  3. redmonda

    redmonda Member

    Messages:
    1,076
    Joined:
    Sep 21, 2015
    Noise is present at the input to the modeler (and thermal noise is present on every component) regardless of whether your guitar is present or not. This noise will be amplified unless it is gated.

    The ISS is not at absolute 0 (Technically, space is near it, but not actually at absolute zero) so there is always a very low level of thermal noise present at the ISS. Whether or not it's significant or not....

    There is also a considerable amount of radiation that hits the ISS which can generate noise.
     
    djd100 and Ron_R like this.
  4. Jay Mitchell

    Jay Mitchell Member

    Messages:
    4,807
    Joined:
    Oct 13, 2007
    That's got to be tongue in cheek.
     
    Aquinas likes this.
  5. Ron_R

    Ron_R Member

    Messages:
    3,389
    Joined:
    Mar 1, 2013
    I read a lot. A lot. One of my closest friend in College's brother designed the heat shielding that protects the space shuttle when it re-enters the atmosphere. He's a smart guy and knows a lot about that kind of thing. We talk.
     
    TonePilot likes this.
  6. MWiggins

    MWiggins Supporting Member

    Messages:
    182
    Joined:
    Jun 15, 2017
    Location:
    Elmwood, CT
    I have had the same thought, I call it post-modeling, makes sense to me.

    With my variaxes and helix I didn't really care how authentic it sounds. In some ways I see all this fuss is like buying a new sports car and asking the dealer to make it look and drive like a model T...
     
    BCy2k likes this.
  7. djd100

    djd100 Member

    Messages:
    2,782
    Joined:
    Nov 25, 2006
    Your thoughts have been addressed correctly by a multitude of people, and calling people a "dick" because your ego tells you that you understand what's been presented to you when it's obvious you don't have a clue, is simply BS!

    Try doing some research on your own seeing you don't believe anyone here.
     
    Aquinas likes this.
  8. djd100

    djd100 Member

    Messages:
    2,782
    Joined:
    Nov 25, 2006
    This of course has nothing to do with tube guitar amps driven into nonlinearity, and sounds like marketing speak to me with regards to HiFi reproduction as well, though I've never tested it LOL!
     
  9. djd100

    djd100 Member

    Messages:
    2,782
    Joined:
    Nov 25, 2006
    I doubt it LOL!
     
  10. br91

    br91 Member

    Messages:
    505
    Joined:
    Dec 24, 2011
    "Many years ago, noted audio designer Bob Carver (Phase Linear, Carver Corp. and others) ran a set of double-blind listening tests to see if "golden eared" audiophiles could be fooled during A/B comparison between tube and transistor amplifiers. Long story short, it turned out that part of the magic formula for emulating a "tube" sound was injecting very small amounts of hum and noise into the program material."

    I wonder if adding a small amount of white noise into a solid state amplifier could be a decent way of adding harmonics to the distortion?
     
  11. Aquinas

    Aquinas Supporting Member

    Messages:
    1,148
    Joined:
    Feb 13, 2005
    Location:
    DFW, for now
    I would hope so, but doubt it. Maybe he just really enjoys comb filters!

    (That was tongue in cheek)

    Ding Ding Ding! We have a winner!

    Somewhere in that answer lies the reason that noise (outside gain-introduced noise floor) is usually included in models. A lot of the character of many amps come from their imperfect power supplies (ripple, hum, crosstalk, parasitic capacitance, and so on). The unfortunate side-effect of modeling that is a bit of added noise - to not add the noise, they'd have to get rid of the beneficial effects (usually harmonic complexity, but many others as well).
     
    TonePilot likes this.
  12. Jay Mitchell

    Jay Mitchell Member

    Messages:
    4,807
    Joined:
    Oct 13, 2007
    It is the burden of proof of anyone claiming that noise "is usually included in models" to establish the truth of that statement. It's easy enough to do: set up your modeler with a processing block in front of the amp - ideally an EQ - that has level control or can be muted. Record the noise with that device set to unity gain, then with it set to max attenuation (ideally completely muted). If noise is actually included in the amp model, you will hear it even with no signal going into the amp block. I have never encountered this in a modeler. If it's there, it's easily enough demonstrated.

    Edit: I just did this with an Amplifirebox. With the gainiest amp sims - Recto and Friedman HBE - the noise completely disappears when you turn on the gate, which is always the first block in the signal chain. Therefore, in the case of these models in this device, there is no noise included the amp models.

    It's always a good idea to test assertions before making them.
     
    Last edited: Mar 14, 2018
    fretworn and djd100 like this.
  13. Aquinas

    Aquinas Supporting Member

    Messages:
    1,148
    Joined:
    Feb 13, 2005
    Location:
    DFW, for now
    Perhaps I wasn't clear - it doesn't have to be "always there noise", but the artifacts thereof, which may only be audible when actually playing...

    That being said, on the Helix and IIRC the Axe-FX, the type of noise you describe (ie, present whether there is input or not) is definitely there. The Helix has the "Hum" and "Ripple" parameters controlling it (or at least some aspects of it, and it does vary from amp to amp), and the Axe has similar parameters. Given the massive difference in overall sound that, say, AC or DC filament heaters can have on overall amp character (mostly through noise/hum characteristics), that it is included is unsurprising to me. Yes, I did just test it just as you say (mute input, not just unplug) on my hardware Helix, out of pure curiosity.
     
  14. Jay Mitchell

    Jay Mitchell Member

    Messages:
    4,807
    Joined:
    Oct 13, 2007
    There have been noise-based schemes implemented in earlier modelers to mask the effects of aliasing. In those cases, broadband noise would be added as signal level increased. That is not a necessity if the levels of aliasing products are kept sufficiently low, nor is it an attempt to duplicate nonideal behaviors of a tube amp, however.

    I don't own a Helix, but I've had the first two generations of Axe-Fx. I never saw evidence that your assertion is correct for those modelers. Neither is convenient right now, but I'll check them for that behavior when time permits.

    There is definitely nonideal noise behavior in modelers, but that is not due to the code in their amp blocks. The inputs of first and second generation Axe-Fx modelers and of the AFB are all microphonic. That's inevitable, however, because capacitors are always microphonic. The front input of my Standard would pick up AM radio stations before I implemented proper termination for the shield lead. The front input of the Axe-Fx II was susceptible to internal clock noise.

    Off the top of my head, the Axe-Fx only has sag and bias controls. Neither of those adds to the noise floor.

    What was the result?
     
  15. Aquinas

    Aquinas Supporting Member

    Messages:
    1,148
    Joined:
    Feb 13, 2005
    Location:
    DFW, for now
    Not what I was talking about - that kind of "dithering" noise-shaping effect isn't really necessary on the newer gens, in any case, as it was the result of poorer DSP performance and band-limiting within the models. I don't think it was that noticeable, even then, and wouldn't have affected the sound in the same way as "modeled amp noise" in any case, as it would have been at the very end of the DSP process, and therefore not interactive with internal gain staging.

    I've never seen evidence that such behavior has any adverse effect on actual usage, as it would always be very, very low level stuff. I've also had no issues with it on either the Axes I've had or the Helix I currently do. Sure, every input section is theoretically microphonic, but the same can be said for any preamplifier of any kind, AFAIK ever. The microphonic tendencies of the modelers are no worse than any tube amp, for sure!

    The Fractal line has them as "Deep Parameters" under "power supply type", according to the Wiki. I remember them being there. Note that this would be extremely low-level stuff, in terms of what is audible without an input signal, unless the parameters in the Axe were cranked high. Helix has it on Page 3 on each and every amp model, as "Hum" and "Ripple". You may not recall them being there, but trust me, Cliff is aware of the effect of power supply interaction and most certainly modeled it!

    As expected - almost unnoticeable at low-to-middle range on "Hum" (though certainly there on any setting above "0"), but above "7" or so quite noticeable. Ripple doesn't have an effect unless there is signal (which is expected behavior), but only becomes readily noticeable when cranked pretty high. You can notice the ripple effect without signal by cranking hum - also expected behavior. I didn't actually measure the noise floor, other than a brief glance at a meter as I was messing about with it, to confirm that the behavior was actually there at lower settings.

    I did try it on a couple of different amps where I was aware of the general behavior (Essex 30 and Cali IV Lead), and they acted generally how you would expect - much more noticeable and interactive on the AC30, more of just a "noise factor" on the Mesa. L6 does seem to have taken into account how each amp operates when modeling these parameters, but I won't claim absolute lack of confirmation bias.
     
  16. Jay Mitchell

    Jay Mitchell Member

    Messages:
    4,807
    Joined:
    Oct 13, 2007
    Correct. I mentioned it for completeness.

    I made no statement to the effect that it did.

    In the case of the three modelers I referenced - AFX Standard, II, and Amplifirebox, the inputs are practically and demonstrably microphonic. It's easily enough experienced: tap on the device housing with an amp block enabled. I referenced this behavior specifically to distinguish it from the intentional coding of nonideal behaviors in the amp block.

    OK, I hadn't been paying attention. That stuff got added recently, long after I had stopped using the AFX II. Regardless, AFX users - and apparently Helix users as well - have the option of turning these parameters off, thereby disabling the modeling of those nonideal behaviors. That fact alone renders the OP moot, as I have been trying to get across.

    If it hurts when you hit your head with a hammer, it's probably a good idea to stop. If you don't like hum and ripple, then would it not be a good idea to set the parameters that cause them to zero?

    It makes no sense to complain about the modeling of things you don't like when you're allowed to turn those things completely off. I realize you aren't doing that. The OP was specifically complaining about hiss, which is not modeled in the amp blocks.
     
  17. Aquinas

    Aquinas Supporting Member

    Messages:
    1,148
    Joined:
    Feb 13, 2005
    Location:
    DFW, for now
    Fair enough, and I must say that I don't know what type of "hiss" the OP was talking about. If it is just the normal induced noise, then there is nothing to be done - as many have said, you gain up a signal enough and you'll always have something there. I can't think of any time where any of the three "component level" modelers actually add anything that can be described fairly as "hiss" to the signal. Hum/Ripple/Power-non-linearities are another thing, but none of them are really "hissy", being mainly present either in the signal itself through modulation (ripple, PS-non-linears) or mostly low frequency (Hum).

    Can't say I noticed it with AxeII's I had - not to say it isn't there. Just tried mucking with the Helix input (out of curiosity), and couldn't get any really noticeable microphonics out of it, even with a high-gain model. Maybe I didn't try hard enough, or maybe they use a slightly different method (different pulldowns, maybe a floating ground, I can think of a half dozen ways to implement the input stage) that reduces the effect. The mic pre was another matter - definitely some very slight microphonics there! I don't really think this is something that actually "matters", I was just curious!

    @TonePilot - what "hiss" exactly were you complaining about, with regard to modellers? Could you describe the effect better? I'm still not quite sure what you were saying they include that they shouldn't...
     
  18. djd100

    djd100 Member

    Messages:
    2,782
    Joined:
    Nov 25, 2006
    Hmmm, modeled power supply induced hum and ripple isn't "modeled noise", rather it's "modeled hum and ripple", where "noise" is all freqs at the same level and "hum" is a specific freq or freqs and their harmonics etc.

    Hum/DC getting in the audio from the power supply can and does effect the audio, but should be properly filtered, especially in correctly designed and operating tube HiFi gear.

    As I understood the OP he was complaining about "hiss", i.e, "noise"?
     
    Last edited: Mar 14, 2018
  19. TonePilot

    TonePilot Supporting Member

    Messages:
    446
    Joined:
    Feb 28, 2018
    Location:
    Canada
    It’s the standard hiss that you hear when you turn up the volume of a tube amp whether or not anything is plugged in. The higher you turn the gain or master, the higher the hiss. This is modeled in say BIAS FX or AmpliTube to the extend I need to add a virtual noise gate.
     
  20. Aquinas

    Aquinas Supporting Member

    Messages:
    1,148
    Joined:
    Feb 13, 2005
    Location:
    DFW, for now
    That's just GAIN, not "modeled hiss"! That is the inevitable result of having a large increase of signal volume (and compression of that). There is nothing "modeled" about it!
     
    Frank Ritchotte, fazer and djd100 like this.

Share This Page