Katana Dev - Audacity science, the T-Comp is a lie.

Started by Beanow, May 23, 2017, 06:48:26 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Beanow

I was pretty confident that the T-Comp parameter does nothing based on how the firmware responded to it. But because it's supposed to be a subtle effect the myth persisted it does something.
Here's my attempt at showing that it doesn't result in any audible changes.

My setup

  • HSS Stratocaster, middle pickup, Texas Special.
  • Line-6 POD HD500X.
    This is my looper. Otherwise it's an empty patch with no modeling, no effects.
  • Boss Katana 100W Combo 1x12 (v1.0.2 firmware).
  • Behringer UCG 102 USB Audio Interface.
    Connected to the Rec Out / Phones of the Katana.

I'm aware both the Pod and Katana have audio interfaces, neither of them work on Linux currently, hence the Behringer.

What I'm looping

Since this is supposed to be some type of compression. I figured dynamics was important.
That's also why I went with the middle pickup. It's clear and punchy, without pickup distortion hiding anything being single coil.

There's a muted strum at the start. This is my sync marker.
Single notes for low output, slowing tempo. Then chord strums.

Both of them I deliberately tried to be inconsistent in case it helps different effects show up.

The test approach

My idea was simple. I will play back the loop through the amp and record the result.
Then when it plays from the start again I will change some settings.

In audio software I will match the starting time of the recordings as close as possible.
For one of the two sections I will invert the polarity of the signal.

If nothing has changed, the two should cancel each other out.
If there was a change, a signal should remain to show it.

Attempt #1, Tweed Amp

For a basic tone, I took the Tweed (default crunch) with ~25% gain and ~60% volume.
EQ and presence all flat. No other effects, no noise filter.

I prepared a compressor (in the Mod slot, before the Amp) with a mild but clearly audible effect. This was off in the base tone, but would be to test my method.

Then I recorded the loop 6 times.
Base > Base > Base > MOD Compressor On/Off > Base > T-Comp sweeps

Then mixed these as follows:
#1 > #2 = Normal
#3 > #4 = Inverted

#1 > #2 = Normal
#5 > #6 = Inverted

This is what came out.


Matching works out.


No change VS toggling the MOD Compressor, success!

Few things to notice. The cancellation is not perfect, probably a mix of analog interference and minute measure and timing differences.
I'm glad I did so many base ones, so it's clear what no change looks like.
The MOD compressor though is very much visible. It goes all the way to max volume, some tiny clipping even.
You can also see that it has an On / Off pattern.

Now the real one.


No change VS T-Comp changes

Nothing. Absolutely nothing to be seen.
I swept across from -20 to +20 multiple times during the second loop.
Using both Fx FloorBoard and my own scripts to try and set the T-Comp value.

Things do not look good for Mr. T-Comp.


Attempt #2, Marshall amp

For the Tweed I had forgotten to note the exact settings.
Besides, the Plexi was mentioned in a report to have audible T-Comp effects.


Other than the amp there are no effects.

The recording order:
Base > Base > Base > T-Comp > Base > Presence & Mid

Since the MOD Compressor last time had a strong effect, this time I decided to try something a little less obvious.
In the final loop I started wiggling the presence a little, then the EQ middle value.

Onto the results:


No change VS T-Comb

I think that settles it. This recording has a lot less noise and cancels out nearly perfectly.
You can't even tell the difference between when the loop is playing or not.
And T-Comp does not -Cut Through- the cancellation at any moment.


No change VS Presence -> EQ Mids
This one is quite interesting. It reveals that subtle *should* show up.

Note we have some cancellation imperfections again in the no change loop.
But in the second loop, the first half has ever-so-slightly more volume. (Highlighted darker)
Note as well that it goes up and down, but does not match the original volume above.
You can see that by sweeping the presence, it let's more or less of the signal through reducing the cancellation effect.

Finally when changing the EQ Mid the effect is actually pretty strong.
As I was moving the knob, I could already hear it had a big effect so I only moved it slightly (between 10 and 2 'o clock). As opposed to the presence which I turned almost fully back and forth.

Conclusion

The T-Comp parameter definitely does nothing.
If you don't believe waveforms, I've exported the original recordings and the canceled mixes, feel free to listen for yourself.

For clarity, all these files in order look like this:


  • TCompTest-Plexi-Full.flac
  • TCompTest-Plexi-PresenceMidCanceled.flac
  • TCompTest-Plexi-TCompCanceled.flac
  • TCompTest-Tweed-Full.flac
  • TCompTest-Tweed-MODCompCanceled.flac
  • TCompTest-Tweed-TCompCanceled.flac


Downloads over IPFS https://ipfs.io/ipfs/QmdTdJ38Yej9AVvbVw5UqPYFWsRA6Hr7gqr1XE112no71f
Mirror on MediaFire http://www.mediafire.com/file/8g4bfo3c2jrxqsc/TCompTest.zip

Beanow

Quote from: snhirsch on May 24, 2017, 05:08:30 AM
Exactly.  I appreciate Robin's efforts to quantify the situation, but am not completely convinced the approach used would necessary flag what T-Comp does.   A better test might be to drive the preamp into clip with a tone-burst generator and watch the leading part of the output waveform with a scope as the T-Comp control is rotated.

If by clip you mean the preamp distortion, both have a very mild crunch. More audible with the tweed.
Anyway, I don't have a generator on hand, would a regular 3xOSC synth with some sawtooths / squares convince you or are there some software generators that you know of?

Feel free to suggest patch settings as well.

admin

#2
T-Comp ( when it works on GP-10 ) makes no difference in clipping  - its a very subtle compression change

gumtown

I haven't tried the "TCOMP" yet, but regardless of whether it works or not,
there are those who truly do believe it does work, so it can stay in the FxFloorBoard editor.

Lets call it the 'placebo effect'.
It works for those who believe in it.   ;)
Free "GR-55 FloorBoard" editor software from https://sourceforge.net/projects/grfloorboard/

vtgearhead

Quote from: admsustainiac on May 24, 2017, 07:40:45 AM
T-Comp ( when it works on GP-10 ) makes no difference in clipping  - its a very subtle compression change

On a GP-10, at least, it's more like a change in slew rate on the attack.  $Deity knows if the Katana is actually doing anything.  I have an HP toneburst generator here, but not a lot of time to spend on the bench at the moment.   

eUphonic

Hello everybody,

Some European senior guitar geek, here.  :D

For my very first post here, I'd like to share a test done with my GP10 while I'm jailed at home because of "C19" (which is sadly not a capacitor).  >:(

Amp used : MS1959. All controls at noon. Cab sim disabled. I think that I've left the gain setting on "low" (not sure). The output setting was on combo amp or something like that.

The attached files below will show you  the effect of the T-comp parameter set @ max (+10) and min (-10), on...

-frequency response;
-phase response;
-THD;
-impulse response (that's where the "compression" effect can be seen as such, in the height of the attack).

As you can see, the difference is subtle.

To put this contribution in perspective, I'll just say that I don't notice more difference than this when I fit my home hot rodded class A 5W amp with a 6V6, a 6L6, an EL34 or any other compatible power tube: tubes don't always do much difference in real life.  ;)

As I've thousands of tests results archived here, I may post other frequency graphs in the next days/ weeks/ months, about the GP10 or about other digital devices mentionned on this forum, if time permits and if it has any interest for other members... In the meantime, keep in mind that my only intention is always to share peacefully some of my findings in order to be useful.


admin

Quote from: eUphonic on April 26, 2020, 01:22:53 AM
Hello everybody,

Some European senior guitar geek, here.  :D

For my very first post here, I'd like to share a test done with my GP10 while I'm jailed at home because of "C19" (which is sadly not a capacitor).  >:(

Amp used : MS1959. All controls at noon. Cab sim disabled. I think that I've left the gain setting on "low" (not sure). The output setting was on combo amp or something like that.

The attached files below will show you  the effect of the T-comp parameter set @ max (+10) and min (-10), on...

-frequency response;
-phase response;
-THD;
-impulse response (that's where the "compression" effect can be seen as such, in the height of the attack).

As you can see, the difference is subtle.

To put this contribution in perspective, I'll just say that I don't notice more difference than this when I fit my home hot rodded class A 5W amp with a 6V6, a 6L6, an EL34 or any other compatible power tube: tubes don't always do much difference in real life.  ;)

As I've thousands of tests results archived here, I may post other frequency graphs in the next days/ weeks/ months, about the GP10 or about other digital devices mentionned on this forum, if time permits and if it has any interest for other members... In the meantime, keep in mind that my only intention is always to share peacefully some of my findings in order to be useful.

Good work

I'm curious about your frequency plotting tools / software / hardware employed?


eUphonic

Quote from: admin on April 26, 2020, 03:08:01 AM
Good work

I'm curious about your frequency plotting tools / software / hardware employed?

In this case (quick test situation), the software is HOLM Impulse and the hardware is a plain old Line6 UX1. :)

gumtown

Thanks for sharing that, the software looks interesting.
Also didn't expect to see that much difference in the T-Comp phase response chart plot.
Free "GR-55 FloorBoard" editor software from https://sourceforge.net/projects/grfloorboard/

vtgearhead

I feel a bit vindicated.  Always felt that T-Comp was doing something :-).

eUphonic

As a footnote, I post below a picture showing the action of the "VR gain" control in a first generation Vox Tonelab (that I've sold for a while now).

If memory serves me, "VR gain" was the master volume of the preamp: cranking it up was causing the onboard 12AX7 to compress.

It's not quite comparable to my T-comp test above, since the test with the Tonelab involved a Marshall JCM800 preamp with all controls at noon, then with Gain AND VR gain full up.

And obviously, I've not used the same frequency analyzer.

But "you'll  get the picture"... :-)