Tone Matching the Internal Cab

White noise is a static sound that has equal energy on every frequency. So has your sine sweep.
Pink noise has a -3dB per octave slope, which makes the spectrum look completely flat on a logarithmic scale.
But it has a slope nevertheless.
So it's normal that white noise and a sine sweep should give the same results.
But you should be good with either white or pink noise.
 
Test Tone Comparison:

Because of the Input from Cliff and MaxTwang i did testing different Testtones:

- Sine-Sweep [synth, third voice, sine, filter 20KHz, frequency by lowest LFO rate] -> TMA: Avg.-Time: PEAK HOLD
- Pink Noise following Cliffs recommendation -> TMA: Avg.-Time: 5.000
- White Noise following Cliffs recommendation -> TMA: Avg.-Time: 5.000

sine_vs_white_vs_pink.jpg


Interpretation: Using White Noise or Sine-Sweep seems to give me identical results: The Matching Curve of two PEQs (Hi and Lo-Cut) in series looks identical frome Sine-Sweep or White Noise as testtone. Matching these two carrier signals against eacht other result in a really flat response. Not so the pink noise .... hmm ...

So, in short: I think in the future i will do it with the white noise! Two reasons: Much more ear-friendly and less work (no need to modify anything to LFO....)

You can only use pink noise if you are doing the "dual-channel FFT" approach. IOW, you must inject the noise into both the DUT and the TMA block simultaneously. Or you must use the pink noise when capturing the original spectrum. You can't use a sweep or white noise and then try to match that using pink noise.

Everyone should really learn how to do the dual-channel approach. It yields far better results than the classic EQ match technique when matching hardware (i.e. amps, preamps, channel strips, etc.).
 
Everyone should really learn how to do the dual-channel approach. It yields far better results than the classic EQ match technique when matching hardware (i.e. amps, preamps, channel strips, etc.).

Cliff, you open my eyes and teach me AGAIN! I´ve done what you say:

pink%20noise%20dual%20FFT.JPG


Result: Perfect Frequency response from the two PEQs (like i used in the tests above).

THANK YOU!

What did i learn today: Measuring flat response don`t have to show a linear, flat curve. But it must be "relative" to the counterpart (local <-> reference). Can i understand it like this?

So, and now ... i have to make a scribble from my personal "learning curve" ;)
 
Last edited:
You can only use pink noise if you are doing the "dual-channel FFT" approach. IOW, you must inject the noise into both the DUT and the TMA block simultaneously. Or you must use the pink noise when capturing the original spectrum. You can't use a sweep or white noise and then try to match that using pink noise.

Everyone should really learn how to do the dual-channel approach. It yields far better results than the classic EQ match technique when matching hardware (i.e. amps, preamps, channel strips, etc.).

By dual-channel approach do you mean splitting the noise/synth signal in the Axe and using the FX Loop to send the noise to the reference device (amp, preamp, channel strip, etc.) at the same time you send the noise to the Amp then TMA blocks (or just TMA if you're not matching an amp or preamp)? That's how I shot these the TMA's below.

These TMA's were shot successively, the only change is the source: pink noise, white noise, sine sweep, guitar (from open E chord to barre chords at the 15th fret). The amp was a Boogie Mark IIB the Axe's model is the USA IIC+ Norm. I used the slave out on the Boogie and put the TMA block after the Amp block (before the Cab) so it is only matching the amp. I set the volume, tone, master, presence and eq on the amp and amp block to the same values (or as close as visually possible on the eq). I was not listening - these setting were purely visual. The pink noise shows how accurate the Axe model is, especially considering I was approximating the EQ settings in the amp block.

TMA_MBIIB_pinkNoise.jpg

TMA_MBIIB_whiteNoise.jpg

TMA_MBIIB_sineSweep.jpg

TMA_MBIIB_guitar.jpg
 
By dual-channel approach do you mean splitting the noise/synth signal in the Axe and using the FX Loop to send the noise to the reference device (amp, preamp, channel strip, etc.) at the same time you send the noise to the Amp then TMA blocks (or just TMA if you're not matching an amp or preamp)? That's how I shot these the TMA's below.

In my understanding: Yes.

These TMA's were shot successively, the only change is the source: pink noise, white noise, sine sweep, guitar (from open E chord to barre chords at the 15th fret). The amp was a Boogie Mark IIB the Axe's model is the USA IIC+ Norm. I used the slave out on the Boogie and put the TMA block after the Amp block (before the Cab) so it is only matching the amp. I set the volume, tone, master, presence and eq on the amp and amp block to the same values (or as close as visually possible on the eq). I was not listening - these setting were purely visual. The pink noise shows how accurate the Axe model is, especially considering I was approximating the EQ settings in the amp block.

Should`nt this result in same Matching-Frequency-Curves? At least between pink / white / sine ? In my tests it does ... as you can see ...

By the way: Is the slave out from the boogie before or after the Power Amp section? And if before, did you switch off the power Amp section of the Axe-Fx block?

If the slave out is before the power Amp and you`ll match the Amp Block WITH power amp ... you have matched apples to pears ... i think ....

From the manual:
"The Slave Level control and Slave Output jack located on the rear panel of the amplifier give
you a direct output feed which faithfully captures the entire sound of the amp and preamp..."
http://www.mesaboogie.com/manuals/Mark IIb.pdf

... so, i understand BEFORE the power amp?!
 
Should`nt this result in same Matching-Frequency-Curves? At least between pink / white / sine ? In my tests it does ... as you can see ...

I've been thinking about this and I think the difference between pink and white noise explains the additional highs in the white noise based TMA. Not sure about the sine sweep results. But the only difference was the noise source.

By the way: Is the slave out from the boogie before or after the Power Amp section? And if before, did you switch off the power Amp section of the Axe-Fx block?

The Slave Out is after the power amp, power amp modeling in the Axe was on.

A/B'ing the Slave Out against the Amp/TMA blocks into the Axe's cab block confirmed the pink noise TMA is the most accurate (slight EQ tweaks in the Amp block and they're indistinguishable). The white noise TMA was also good but a little brighter. The sine sweep and guitar based TMA's have too much mid-range 'honk' which is a problem I've had trying to match this amp in the past.

This has been a really tough amp to match (and profile). Last night I finally tried matching the Slave Out and deal with the speaker by shooting an IR. I also figured I'd try white and pink noise.
 
You shouldn't be getting much difference using the various techniques. The whole key is to capture BOTH the reference and user spectra SIMULTANEOUSLY. Don't capture one then the other. Start both by pressing X and then Y immediately after. Stop and acquire by pressing Enter.

That said, if you do not have the gain the same between the model and the amp you will get different results, especially when using noise signals. You need to refine the gain by switching between the amp and the model. Technically the best way to match the gain is to use noise and vary it's level and compare the compression but that requires a spectrum analyzer.

Once you have the gain matched, then do another match.

I did a presentation at Axe-Fest on how to match an amp. I don't know what's taking so long on those videos.
 
I did a presentation at Axe-Fest on how to match an amp. I don't know what's taking so long on those videos.

oh man .... yeah, where are those videos .... i bet there are thousends of people waiting for those .... !!! Me too! Hope we`ll see them soon .... ;)
 
What did i learn today: Measuring flat response don`t have to show a linear, flat curve. But it must be "relative" to the counterpart (local <-> reference). Can i understand it like this?

So, and now ... i have to make a scribble from my personal "learning curve" ;)

Yes, it's all about the "relative" difference. The TMA block is a difference analyzer. It looks for the difference between the two spectra. If you feed the EXACT same signal to both SIMULTANEOUSLY then you are guaranteed to only measure the difference since you've removed any dependence on the stimulus.

This is why matching an amp or other hardware is so much easier than matching a recording. When you match a recording you get a time-averaged spectrum for the recording. Then you play your guitar and create a time-averaged spectrum for it. Neither of those spectra are very representative of the frequency response of the respective systems since they depend on the stimulus. The difference between them is therefore not very accurate because the stimulus was different.

However, if you send the exact same stimulus to, for example, an amp and a model of that amp simultaneously then it doesn't really matter since your are only looking for the difference.
 
If you feed the EXACT same signal to both SIMULTANEOUSLY then you are guaranteed to only measure the difference since you've removed any dependence on the stimulus

Yep, got that! Once i played just for fun guitar simultaneously into local & reference - matched it and because both frequency responses were identical the match was a flat line :)

Appreciate, if you`ll find time to post your PowerPoint Cliff! Many thanks!!! by the way: I am very happy, that you create this awesome machine!!! And i`m very lucky, because it gets better and better ... can`t wait for the next step ... and try out the Fuchs and the learn some John Mayer licks using the Two Rock ;)
 
Later on I will dig out my PowerPoint presentation on amp matching and see if I can figure out how to publish it to the web.

i know Keynote (Apple's version of Powerpoint) can export presentations as a video, so you could then put it on youtube. you set an interval between slides/actions and wind up with a 5 minute or whatever vid. i could do that for you if you want. keynote should be able to open up powerpoint presentations. unless powerpoint can do that too :)
 
You shouldn't be getting much difference using the various techniques. The whole key is to capture BOTH the reference and user spectra SIMULTANEOUSLY. Don't capture one then the other. Start both by pressing X and then Y immediately after. Stop and acquire by pressing Enter.

That said, if you do not have the gain the same between the model and the amp you will get different results, especially when using noise signals. You need to refine the gain by switching between the amp and the model. Technically the best way to match the gain is to use noise and vary it's level and compare the compression but that requires a spectrum analyzer.

Once you have the gain matched, then do another match.

I did a presentation at Axe-Fest on how to match an amp. I don't know what's taking so long on those videos.

Thanks for the info, makes sense. That's probably the root of the difficulties I've had matching this amp. At least I now have results that are more fun than the process of trying to get more accurate (for now).
 
The PP presentation includes a possible solution on how to organise the various amps and categorise them in the 3 (4) basic categories in the Axe. This and 'sort by name' would be very effective...
 
Back
Top Bottom