Waves SSL E-Channel for just $29!

[poshook]

er... not to burst any bubbles, but... did you spot the mistake?



poshook - can you please measure again with the following?

• a -18dBFS sine signal?
• the compressor engaged but not compressing (read: threshold as high as possible, ratio at 1:1, no gain compensation)?
• post the difference of one EQ band boosted (one plot) and one not (a separate plot)? (so two plots - let's say: 3kHz, boost about 6dB)
• maybe even use a different FFT?

I'm pretty sure the results would look vastly different...

Also... why do you compare a NEVE console by Brainworx with an SSL console by Waves?

I compared with BX Console because some1 in this thread mentioned this one as a better more analog alternative with non-linearities. I know the difference (Neve design vs SSL) but in terms of non-linearities the BX_Console is probably the most cleanest strip on the market. Filters/EQ/Dynamics are just curves. The only section making this plugin special is TMT which is just separate section working over the top of the other sections and alternate some attributes.

[Compyfox]

Since I don't have bx_console, and I didn't want to waste my demo time as fan of NEVE consoles, I can neither confirm nor deny what you measured in this case.

But regarding Slate Digital and testing compressors... you should definitely re-measure again.

[poshook]

Since I don't have bx_console, and I didn't want to waste my demo time as fan of NEVE consoles, I can neither confirm nor deny what you measured in this case.

But regarding Slate Digital and testing compressors... you should definitely re-measure again.

Understand. About Console. Regarding Brainworx CEO VXS console is Neve but very clean which is not the case of vintage designs. The signal must be pretty hot to reach the saturation. Sound of saturation is not a big deal (very harsh and bad sounding) as the console is not designed to be a saturation machine like its sisters from the past.

[Ichad.c]

Testing a compressor for THD while compressing isn't a test for analog modelling - a purely digital comp will do it too. The hart of a compressor in a nutshell is: audio signal -> rectification -> filtering(this is the control voltage). Rectification is a highly non-linear process, if you are a guitar player you'll know this well, it is how the classic octave up pedals create their sound. The filtering in compressors turn this "octave up" rectification and multiplies it with your signal - that is compression. Depending on the attack/release times(filters) the harmonics will vary, general rule -> the faster the time, the less filtering and that results in more harmonics on the output signal.

A better test for analog modelling(for filters not comps) is too test if the transformers are modeled or not. First test 1kHz for distortion, then test 50Hz, if the transformers are modeled, there will be more distortion @50Hz.

FYI, most classic EQs(even Neve) are designed to be pretty clean, with the exception of inductor based EQs, most of the distortion comes from gain make-up stages and transformers.

[lkjb]

Ugh... "null down to -80dBFS"... oh please!

That's not a null test at all if most console sim's crosstalk kicks in at -76dBFS. If it's below -110dBFS - then we start talking, and even then, it is not(!) a "null test". I can "null" with any digital EQ with this...tolerance as well.

I probably totally misunderstood what you meant to say here but you are aware that if you have two identical tracks @ 0dBFS peak and add them inverted with one track's volume at -0.00087 dB you will get a remaining signal of about -80 dBFS. So -80 dBFS is pretty much a successful null test regarding that no one will be able to hear a difference.

[BRBWaffles]

Ugh... "null down to -80dBFS"... oh please!

That's not a null test at all if most console sim's crosstalk kicks in at -76dBFS. If it's below -110dBFS - then we start talking, and even then, it is not(!) a "null test". I can "null" with any digital EQ with this...tolerance as well.

I probably totally misunderstood what you meant to say here but you are aware that if you have two identical tracks @ 0dBFS peak and add them inverted with one track's volume at -0.00087 dB you will get a remaining signal of about -80 dBFS. So -80 dBFS is pretty much a successful null test regarding that no one will be able to hear a difference.

I too got the distinct impression that he did not know what the hell he was talking about, either that or the profundity of his insight is beyond the scope of my comprehension.

[poshook]

Ugh... "null down to -80dBFS"... oh please!

That's not a null test at all if most console sim's crosstalk kicks in at -76dBFS. If it's below -110dBFS - then we start talking, and even then, it is not(!) a "null test". I can "null" with any digital EQ with this...tolerance as well.

I probably totally misunderstood what you meant to say here but you are aware that if you have two identical tracks @ 0dBFS peak and add them inverted with one track's volume at -0.00087 dB you will get a remaining signal of about -80 dBFS. So -80 dBFS is pretty much a successful null test regarding that no one will be able to hear a difference.

Could that difference be more pronounced on e.g. 50 tracks together? Just thinking? Technically I have no idea

[BRBWaffles]

Ugh... "null down to -80dBFS"... oh please!

That's not a null test at all if most console sim's crosstalk kicks in at -76dBFS. If it's below -110dBFS - then we start talking, and even then, it is not(!) a "null test". I can "null" with any digital EQ with this...tolerance as well.

I probably totally misunderstood what you meant to say here but you are aware that if you have two identical tracks @ 0dBFS peak and add them inverted with one track's volume at -0.00087 dB you will get a remaining signal of about -80 dBFS. So -80 dBFS is pretty much a successful null test regarding that no one will be able to hear a difference.

Could that difference be more pronounced on e.g. 50 tracks together? Just thinking? Technically I have no idea

I mean, yeah. 50 tracks together would create quite a sum of additional noise. But you have to think that it's -80dB for every track individually, and all those tracks could be at 0dB. In the end it likely doesn't sum the way you might intuit it to. Even if it did I don't think a statement like "you need to use this plugin on at least 50 tracks in order to hear a difference against our free competitors" is a very good selling point. At the end of the day I would not give Waves money when there are tools like the CS-3301 that do the same thing with with a dozen additional features, or free options like the DD Channel that pretty much do everything the same at no charge. If you're willing to buy the SSL E-Channel, knowing what you know about it, the only reason I could comprehend would be brand loyalty and not much else.

[poshook]

Ugh... "null down to -80dBFS"... oh please!

That's not a null test at all if most console sim's crosstalk kicks in at -76dBFS. If it's below -110dBFS - then we start talking, and even then, it is not(!) a "null test". I can "null" with any digital EQ with this...tolerance as well.

I probably totally misunderstood what you meant to say here but you are aware that if you have two identical tracks @ 0dBFS peak and add them inverted with one track's volume at -0.00087 dB you will get a remaining signal of about -80 dBFS. So -80 dBFS is pretty much a successful null test regarding that no one will be able to hear a difference.

Could that difference be more pronounced on e.g. 50 tracks together? Just thinking? Technically I have no idea

I mean, yeah. 50 tracks together would create quite a sum of additional noise. But you have to think that it's -80dB for every track individually, and all those tracks could be at 0dB. In the end it likely doesn't sum the way you might intuit it to. Even if it did I don't think a statement like "you need to use this plugin on at least 50 tracks in order to hear a difference against our free competitors" is a very good selling point. At the end of the day I would not give Waves money when there are tools like the CS-3301 that do the same thing with with a dozen additional features, or free options like the DD Channel that pretty much do everything the same at no charge. If you're willing to buy the SSL E-Channel, knowing what you know about it, the only reason I could comprehend would be brand loyalty and not much else.

For me 80% is workflow and with E-Channel I always find sweet spot very fast. I like how it looks including placement of the GUI elements, and all those SC capabilities. 20% is the sound which I like and know for many years. So everything is where it should be and sounds as it should. All you mentioned alternatives lack DYN-SC, DD is only for windows. I am on Mac

[Compyfox]

Testing a compressor for THD while compressing isn't a test for analog modelling - a purely digital comp will do it too. The hart of a compressor in a nutshell is: audio signal -> rectification -> filtering(this is the control voltage). Rectification is a highly non-linear process, if you are a guitar player you'll know this well, it is how the classic octave up pedals create their sound. The filtering in compressors turn this "octave up" rectification and multiplies it with your signal - that is compression. Depending on the attack/release times(filters) the harmonics will vary, general rule -> the faster the time, the less filtering and that results in more harmonics on the output signal.

I honestly did not expect the answer to the "riddle" I brought up so soon.

I think in combination with my post about the measure request, people should now know "how to do it properly". Then again, that's just the THD+N test.

A better test for analog modelling(for filters not comps) is too test if the transformers are modeled or not. First test 1kHz for distortion, then test 50Hz, if the transformers are modeled, there will be more distortion @50Hz.

IMO 50Hz in plugin form usually gives no suitable results, but that doesn't mean that you can't measure it regardless.

FYI, most classic EQs(even Neve) are designed to be pretty clean, with the exception of inductor based EQs, most of the distortion comes from gain make-up stages and transformers.

Since we can't measure this in plugin form, because you can't measure individual blocks of the circuit (unlike hardware), we can only go with "what goes in, and what goes out".


Either way, thanks for the contribution to this thread.
This is adding a lot to the knowledge base regarding "measurements".

Ugh... "null down to -80dBFS"... oh please!

That's not a null test at all if most console sim's crosstalk kicks in at -76dBFS. If it's below -110dBFS - then we start talking, and even then, it is not(!) a "null test". I can "null" with any digital EQ with this...tolerance as well.

I probably totally misunderstood what you meant to say here but you are aware that if you have two identical tracks @ 0dBFS peak and add them inverted with one track's volume at -0.00087 dB you will get a remaining signal of about -80 dBFS. So -80 dBFS is pretty much a successful null test regarding that no one will be able to hear a difference.

I too got the distinct impression that he did not know what the hell he was talking about, either that or the profundity of his insight is beyond the scope of my comprehension.

Sorry but now you're confusing me, lkjb.



It's simple really: we have two plugins that we want to compare with each other. In this case, it was Waves SSL E-Channel vs TBProAudio CS-3301. The premise is to "null" the signals with each other, with the same or similar filter settings. Or in other words, completely phase invert.


I've yet to do this myself with this plugin combination, but let's remain theoretical:

Let us assume that neither of these plugins introduce saturation or additional noise - best case if we can turn that off. We only focus on the filter curves itself. Assuming we want to "null" a boost of 250Hz with +6dB, a cut at 1kHz with -6dB and another boost at 10kHz at +10dB. We setup the source (in this case the Waves Plugin) to that value, then try to get as close as possible with the target (in this case TBProAudio or whatever EQ you prefer)... you then proceed to phase invert either of these plugins (let's say the "source") and see what happens on the output meter.


Now... if there is no modulation/channel shift (independent channel handling)/whatever involved, the lower the readout on the group or summing bus, the better.

A "null" would mean, that the signal is completely phase inverted and you will get a readout of -inf

This is not the case here (!!!), we talk about -80dBFS. And that is far from being a "null". You will still hear glitches - albeit faint ones. Which might be worse, if the plugins introduce crosstalk (stereo channels!), or if channels are treated individually rather than globally. A "null" at -110dBFS is definitely a better result, but you will still have the artifacts.

So no, this is not(!!!) a "null test" - this is testing how close you can get. And we're still disregarding as to what settings are on the target plugin, and how much they offset compared to the filter behavior we wanted to recreate with the source.

If it would completely "null out", then hooray - it's the "most standardized, 1:1 filter behavior you can found on the market" (read: the same code base was used behind the scenes, which is mostly not the case unless it was an open source project that only altered the UI). This is the reason why a "null test" is always headache inducing... it might put your mind at ease after seeing "oh, it's nothing special to me - it nearly works the same", but it's useless on the long run. In most cases, the result will not(!) giving any leverage regarding the topic "this is nothing but a mere digital EQ".


Have I made myself clear now?

[pc2000]

Maybe the title of this thread should be changed to reflect the attack on a plugin that's just one out of thousands-in plugin land. What makes it particularly ripe for attack? Overloud for example is charging $119 for the EQ495, which sounds awful. It has zero so called mojo, with loads of freeware being far better. One thing I can say about Waves plugins... is that I haven't experienced a single crash or glitch in any DAW I've used them in.

Luv or hate their offerings.... they seem rather solid even if the code is dated in number of them? I'm not convinced that hardware can be fully modeled without compromises. I only look at FX plugins as being inspired by designs. To expect an actual model is to believe in Santa!

[poshook]

So nice this thread is going to be...

[Compyfox]

Add to that, that Waves keeps their code updated/alive even up until this day (L1 and MaxxBass anyone?)... well, yeah.

[poshook]

Add to that, that Waves keeps their code updated/alive even up until this day (L1 and MaxxBass anyone?)... well, yeah.

MaxxBass here

[BRBWaffles]

It's simple really: we have two plugins that we want to compare with each other. In this case, it was Waves SSL E-Channel vs TBProAudio CS-3301. The premise is to "null" the signals with each other, with the same or similar filter settings. Or in other words, completely phase invert.


I've yet to do this myself with this plugin combination, but let's remain theoretical:

Let us assume that neither of these plugins introduce saturation or additional noise - best case if we can turn that off. We only focus on the filter curves itself. Assuming we want to "null" a boost of 250Hz with +6dB, a cut at 1kHz with -6dB and another boost at 10kHz at +10dB. We setup the source (in this case the Waves Plugin) to that value, then try to get as close as possible with the target (in this case TBProAudio or whatever EQ you prefer)... you then proceed to phase invert either of these plugins (let's say the "source") and see what happens on the output meter.


Now... if there is no modulation/channel shift (independent channel handling)/whatever involved, the lower the readout on the group or summing bus, the better.

A "null" would mean, that the signal is completely phase inverted and you will get a readout of -inf

This is not the case here (!!!), we talk about -80dBFS. And that is far from being a "null". You will still hear glitches - albeit faint ones. Which might be worse, if the plugins introduce crosstalk (stereo channels!), or if channels are treated individually rather than globally. A "null" at -110dBFS is definitely a better result, but you will still have the artifacts.

So no, this is not(!!!) a "null test" - this is testing how close you can get. And we're still disregarding as to what settings are on the target plugin, and how much they offset compared to the filter behavior we wanted to recreate with the source.

If it would completely "null out", then hooray - it's the "most standardized, 1:1 filter behavior you can found on the market" (read: the same code base was used behind the scenes, which is mostly not the case unless it was an open source project that only altered the UI). This is the reason why a "null test" is always headache inducing... it might put your mind at ease after seeing "oh, it's nothing special to me - it nearly works the same", but it's useless on the long run. In most cases, the result will not(!) giving any leverage regarding the topic "this is nothing but a mere digital EQ".


Have I made myself clear now?

Nobody is claiming that a 100% null is the only relevant or valid criteria upon which to base a comparison. The argument is whether or not a plugin can be cancelled beyond the point of a perceptible difference. This is because this is more or less the criteria people use to differentiate options when they make a plugin purchase to begin with. If I can create a sufficient polarity inversion between the SSL E-Channel and the CS-3301, or even ReaEQ for that matter, so that the resulting signal is -80dB, "this is nothing but a mere digital EQ" doesn't seem like an entirely unreasonable assessment anymore.

Do me a favour. Record some pink noise, normalize the track to 0dB, copy the track, play the two tracks, and adjust your master volume to a tolerable listening level. Now, reverse the polarity on one and adjust one fader to -0.01dB. The resulting signal should be around -70dB or so. Without readjusting your master fader, tell me with a straight face that the difference is worth $100. How about $50? How about $20? How about $1? How about a penny, in all seriousness?