Helix Talk

Can some please invite DieSchmalle over here, I can't PM him on TGP.

Yes, the Elmsley NFB control is totally broken, and behaves nothing like a real NFB resistor mod on an amp.
Moreover, a NFB mod on a real amp will boost the impedance curve frequencies as the NFB goes down and you will see/hear a very clear impedance curve shaped eq boost when the NFB is open/disabled, nothing like that happens with the Elmsley.

When the Elmsley model is biased towards Class A, Pres & Depth at 0, with NFB at exactly 0.38 the amp goes bananas and dips 50dB, and the NFB control in general is nothing like a real NFB control.


Elmsley NFB.png


Elmsley.png
 
Last edited:
Welcome DieSchmalle!
I've been trying to PM you on TGP for ages.

All the cool kids are here, plus there is more Helix discussion here than on TGP since the great NDSP purge last year.
 
James Freeman said:
Can some please invite DieSchmalle over here, I can't PM him on TGP.

Yes, the Elmsley NFB control is totally broken, and behaves nothing like a real NFB resistor mod on an amp.
Moreover, a NFB mod on a real amp will boost the impedance curve frequencies as the NFB goes down and you will see/hear a very clear impedance curve shaped eq boost when the NFB is open/disabled, nothing like that happens with the Elmsley.

When the Elmsley model is biased towards Class A, Pres & Depth at 0, with NFB at exactly 0.38 the amp goes bananas and dips 50dB, and the NFB control in general is nothing like a real NFB control.


View attachment 6788

View attachment 6787
Click to expand...

Maybe someone should tell Line 6 that NFB doesn't mean No fu***ng Bass!
 
!!!Nerd stuff disclaimer - read at your own risk!!!
James Freeman said:
[...]Yes, the Elmsley NFB control is totally broken, and behaves nothing like a real NFB resistor mod on an amp.[...]
Click to expand...
Or does it?
Keep in mind that Sam (the creator) said it's set on the edge of explosion. I read this as "max plate voltage that he could get away with" - a state where a real power tube's plate is fried in a short time. If you add bias current the plate dissipation increases - but the (direct) current through the tube is saturated and can't increase anymore. I think this leads to async distortion because the upper half of the signal waves now hit the ceiling.

That signal is then influenced by the speaker which leads to an impedance curve imprint (boosted levels around speaker resonance and voice coil increment in the high frequencies ) - but why that low mid hump so exaggerated?
Well - this signal is then send back to the phase inverter (the NFB parameter sets the amount) and subtracted from the preamp signal.
This means a low-mid boosted signal is fed back into the power amp. That boosted lower mids are hitting said hard power amp ceiling first resulting in a low-mid reduction after the power tubes / in the speaker signal.

That's my 2 cents, I could be wrong of course - but at least we have a hypothesis.
 
Last edited:
DieSchmalle said:
!!!Nerd stuff disclaimer - read at your own risk!!!

Or does it?
Keep in mind that Sam (the creator) said it's set on the edge of explosion. I read this as "max plate voltage that he could get away with" - a state where a real power tube's plate is fried in a short time. If you add bias current the plate dissipation increases - but the (direct) current through the tube is saturated and can't increase anymore. I think this leads to async distortion because the upper half of the signal waves now hit the ceiling.

That signal is then influenced by the speaker which leads to an impedance curve imprint (boosted levels around speaker resonance and voice coil increment in the high frequencies ) - but why that low mid hump so exaggerated?
Well - this signal is then send back to the phase inverter (the NFB parameter sets the amount) and subtracted from the preamp signal.
This means a low-mid boosted signal is fed back into the power amp. That boosted lower mids are hitting said hard power amp ceiling first resulting in a low-mid reduction after the power tubes / in the speaker signal.

That's my 2 cents, I could be wrong of course - but at least we have a hypothesis.
Click to expand...

With NFB at zero, colder Bias and lower Master volume I should see a very distinct impedance curve boost, every output transformer coupled tube amp does that even at freezing cold Class B, but I don't see an impedance curve imprint at all with the Elmsley, it's flat line.
Feeding back of this impedance EQ curve is what makes the Presence and Depth controls on a real amp operate and sound like they do, but on the Elmsley they barely have any range with colder bias, there is no impedance curve imprint, it's the same before the power tubes and after the transformer, it's a flat line.
I've seen the same behavior with my Resistive (flat) load box, the Presence and Depth controls have much smaller range even with a lot of NFB compared to a Reactive load or a real cab.
This also tells me they don't have a virtual impedance curve nailed yet, even faking an impedance curve by feeding back a simple impedance shaped EQ to the phase inverter will operate much more realistically than a flat line.

Anyway, this one is a no from me dawg, I much prefer their measured amp models which operate like tube amps should.
 
Ok, since my last post was kind of a wild guess which made me feel uncomfortable - I sat down for some minutes and measured the Elmsley's frequency response.

Measurements
These preamp settings gave me a flat response to then measure the full amp:
Gain 1.0
Bass 5.0
Mids 6.6
Treble 0.0

Measurements with Depth 0.0, Presence 0.0, Bias 5.0 and NFC (Negative Feedback Control) as parameter:
Elmsley NFC.png

We see a controlled flat response with NFC = 0.0 as expected from an amp with high negative feedback.
We see a strong impedance imprint with NFC =1.0 as expected from an amp with no negative feedback.

Context
From the release notes:
"The other new key feature of The Elmsley's power amp section is the Negative Feedback (NFB) parameter, where the amount of feedback within the power section's response can be dialed anywhere from wild and unhinged, tight and punchy, and anything in between. As with many amplifier designs, the presence and depth controls are part of this negative feedback circuit and functions to control the brilliance and resonance of the power amp, respectively. As the NFB param is reduced, the presence and depth controls' influences are also influenced, and they become effectively deactivated when the NFB knob is at zero. These interactions allow the player to create countless tonal variations to suit their needs.”

Result
This is puzzling. Depth / Presence are controls that partially reduce the negative feedback on certain frequencies - so according to my measurements and conclusions they shouldn't work with NFC at 1.0 - not at 0.0 as described in the release notes.
Or in other words: The NFC (NFB parameter) seems to have a bug - it looks flipped (1.0 should be 0.0 and 0.0 should be 1.0).
 
Last edited:
DieSchmalle said:
We see a strong impedance imprint with NFC =1.0 as expected from an amp with no negative feedback.
Click to expand...
Try to measure with Bias at 2.
The impedance imprint should be even stronger with cold Class B bias and no NFB, but it is non-existent.
I think there's more than one bug responsible for what we're seeing.

This is how working Helix amp models measure, very distinct impedance bump at around 100Hz and Presence control sweep.

brit-2204-png.6529


placater-png.6530
 
DieSchmalle said:
Elmsley NFC.png

We see a controlled flat response with NFC = 0.0 as expected from an amp with high negative feedback.
We see a strong impedance imprint with NFC =1.0 as expected from an amp with no negative feedback.
Click to expand...
I disagree, this is not an impedance curve imprint, this is a weird phase anomaly.
The two pictures I posted above are a good representation of what an impedance imprint looks like.

Is the NFB going back to the PI or earlier in the preamp?
Sam talks about "distinct saturation characteristics across the lower and upper registers." so maybe there is a crossover or a split parallel paths in that 100Hz region and the NFB connected before that?
 
Last edited:
James Freeman said:
I disagree.
This is not an impedance curve imprint, this is a weird phase anomaly.

Is the NFB going back to the PI or earlier in the preamp?
Sam talks about "distinct saturation characteristics across the lower and upper registers." so maybe there is a crossover or a split parallel paths in that 100Hz region and the NFB connected before that?
Click to expand...
Not a regular imprint but everything above 150Hz looks like a voice coil's effect on a (current driving) tube power amp (comparing your red curves with my NFC=1.0).
Below 150Hz there is a different thing going on - it's not a regular speaker resonance.

In sum this looks like a mixed signal indeed - like two power amps where NFB only affects one them.
 
Last edited:
Being a retired EE with a fair professional background in analog circuits I should probably know much more about this than I do. All I can say about the Elmsley is that ignorance is bliss and in my relatively brief experimentation with it I liked it most with the NFB set at zero or close to it, at least in the context of the preset I dropped it into. When I played around with it I didn't even know what NFB stood for. Interesting I preferred the generally flatter response. I'll have to revisit it in a mix-like setting.
 
DieSchmalle said:
What Drive and Master settings did you use there?
Click to expand...
Input sine sweep is at -50dB, Drive is at 2, Master at 2, output is crystal clean and volume boosted afterwards.
I also used an AutoEQ to make the Preamp response flat, then measured the full Amp to get just the power amp response.
It's not my first time.

Anyway, something is up with the key feature of this model and the likelihood that Line 6 will fix that is nill.
First attempt at NFB.... Failed. :cry:
 
I hate to be the guy that calls out bad models all the time, I only wish the best for Line 6 and Helix, this is my modeler of choice as you know.
But I also feel it's my duty to report this, how else would they improve?
I don't think a modeler company can grow and improve if the customer feedback is nothing but kissing ass. :idk
Fuck it, I don't care if I get some hate on forums, I'd rather see an honest company grow and improve for the better good of everyone in the future.
 
Last edited:
You must log in or register to reply here.

Similar threads

gearJunkie
Guilty of modeler neglect
2
Replies
22
Views
2K
Mongillo19
M
Mikael Dez
Need advice (if this is even possible - controlling HX Stomp with Tonex via MIDI)
Replies
0
Views
124
Mikael Dez
Mikael Dez
wharf rat
Helix Floor @ $1399 or HX Stomp + Pedals
Replies
5
Views
495
JCW308
JCW308
Alex Kenivel
Bonex One vs Nana Cotex vs Why Bother?
Replies
17
Views
1K
Alex Kenivel
Alex Kenivel
Mikael Dez
Neural DSP Talk
Replies
19
Views
3K
Jarick
Jarick
Back
Top