do all LAZ EQN have an 18kHz notch?

I've had these units for years, built two myself and bought two off of Marcocet and I'm just getting around to putting these back into service. Funny thing is I remember noticing a really intense HF weirdness when boosting both the mid and high bands. I finally fuzzmeasured it and was rewarded with these:

This is a graph of all the mid bands with full boost and low Q.
https://postimg.cc/image/fz6927uhp/

And here we have full boost low Q at 9kHz and full boost 25kHz shelf.
https://postimg.cc/image/p6yhix99p/

And finally the unit in bypass, free air!
https://postimg.cc/image/v7w6fzggd/

What could be creating this 18kHz notch? The secondary frequency 'hump' in the midband becomes more pronounced as the frequency increases, but I assume a vintage 1084 does not do this?  The two pairs of units were built with two different BOM part numbers by two different people yet exhibit this effect so I assume this is a universal 'feature' of the EQN. It seems like something is sucking out all the 18kHz above the 1.6kHz setting. Any suggestions? Anyone that can corroborate?

For a start, do a frequency response curve with the eq set flat. I think you will find its not a notch, its flat compared with the lf, but with eq boosts applied its hard to see whats really going on.

Certainly a good idea. I've attached a plot of the EQ in with all bands off and it looks pretty flat. This is all measuring at 192kHz just to be clear. I guess to be more precise, what I mean is that something is not allowing 18kHz to be boosted in the mid band, the rest of the bell curve tracks fine above and below that particular frequency

paranoidandroid said:

..... something is not allowing 18kHz to be boosted in the mid band, .....

Click to expand...

What ? 18kHz in the "mid-band" ?  BTW, can you hear this notch ?

The notch becomes very apparent when you try to boost 3.2 - 9kHz in the midband (which is the highest frequency available) and also boost the high shelf. As you can see in the second image (LAZ1084 double boost) from my first post.

The original 1084 had a 16kHz shelf but Peter put in both a 20 and 25kHz air band option in the EQN. This is an awesome feature and has allowed me to not need a Maag EQ! except that the mid band interacts to make this notch at high mid frequency options.

I'm youngish so I don't know if this 20-60kHz boost is a design oversight/limitation of the original 1084 and wasn't an issue because of fidelity restrictions of the time or if this is an issue unique to the LAZ EQN adaptation.

One thing that was dropped from the LAZ EQN is the LPF which, look at that, has an 18kHz option as it's highest setting...

.... this 20-60kHz boost is....

Click to expand...

....nothing else than expensive cerebral m*********** (self-censored to become political correct)

Where is music played mostly ?  Exactly, on the radio (beside sh*tty youtube, smartphones etc.)

Did you know that at FM-Radio the bandwith is limited very sharply to 15kHz ?

So how much of this (inaudible) 20-60kHz boost do you think will be measurable ?

You can say now, yes, but on CD.....

Ok, the sampling rate of a CD is 44,1 kHz.  The sampling rate has to be twice the maximum frequency.  So the highest possible frequency on CD is 22kHz.  Now, how much of this (inaudible) 20-60kHz boost do you think will be measurable ?

As an engineer I'm often delivering files at 96/24 where 48kHz very much exists and I'm aware that almost all of that stuff gets chopped right off when a streaming service converts to mp3 or ogg or whatever. It is however my job to at least be aware if super-harmonic content is all over my deliverables as who knows what electrostatic listener is going to have a horrible time with the HD96/24 download he paid for. 
But that's not really the issue here, I'm more concerned about curves.

What is bothering me is that I can't get a smooth shelf boost while also boosting the mid band at high frequencies. This is The Lab, I can't think of a better place to expensively masturbate regarding the fine details of an electronic circuit. If you're familiar with the Maag EQ you'll understand that having a shelf start at 25kHz  or higher is more useful because it moves the gentlest part of the curve into the top of the audible spectrum. The real steep stuff is up out of the audible band. The LAZ EQN shelf does that really well! But then this mid band comes in and puts a seemingly unnecessary dip right in that slope.

All I want to know is, is there a way to remove this dip so we can have a proper additive effect when using both bands? Or is this just a side effect of inductive EQs? On a lead vocal or drum kit for instance, this sort of dip is totally audible and unfortunately ruins the EQN for this sort of air band use.

This is The Lab, I can't think of a better place to expensively masturbate regarding the fine details of an electronic circuit.

Click to expand...

I think here is a better place:
https://groupdiy.com/index.php?topic=37604.0
or here:
http://www.groupdiy.com/index.php?topic=36096.0

..... If you're familiar with the Maag EQ you'll understand that having a shelf start at 25kHz  or higher....

Click to expand...

Sorry no, I can't be familiar with every overprized audiofool-tool, but I had a look at their homepage:

The Maag Audio EQ4® (500 Series) is a one channel 6 band equalizer with AIR BAND® (shelf boost from 2.5 to 40kHz), ....

Click to expand...

Now, can you tell me if there is a significant difference between 2,5 kHz and 25 kHz ?

BTW, I bet that the secret "AIR BAND ®" boost is nothing else than something like an aphex exciter.... you should try one - or one of the countless clones - for comparision.

analogguru said:

Sorry no, I can't be familiar with every overprized audiofool-tool, but I had a look at their homepage:Now, can you tell me if there is a significant difference between 2,5 kHz and 25 kHz ?

Click to expand...

Are you absolutely sure you want to be participating in this thread? Do you own a EQN? Do you have a 1084 you'd like to test out to make a comparison? Can you explain why the inductor might be cutting off at 18kHz or how to mod the circuit to extend the bandwidth?

Is there anything at all that you have to contribute from a basis of prior knowledge? Or are you here for some other reason?

Interesting problem!  Can you post a schematic?  I thought I had it on my computer but can't find it.

This is the original that Peter posted a long time ago which to me is very difficult to read, but maybe your eyes are better than mine!

Aha and here is from the publicly available users manual. BA 211 is midband. You can see which sections he cut out, aside from the mic amp, BA194 the LPF section:

So in your first picture you are boosting mid and highs at the same time? What you are seeing is not a notch at 18K, its because you are boosting two bands, and the 18K is not covered by either of the bands. Time for you to read up about eq's.

radardoug said:

So in your first picture you are boosting mid and highs at the same time? What you are seeing is not a notch at 18K, its because you are boosting two bands, and the 18K is not covered by either of the bands. Time for you to read up about eq's.

Click to expand...

The very first picture 'LAZ1084 mid boost' is actually just the mid band boosted, at each frequency, from 1.6kHz to 9kHz. The image 'LAZ1084 double boost' is that 9kHz boost with the 25kHz shelf fully up as well, so that's the image with both bands boosted.

This could be due to parasitic capacitance in the EQ inductor. I noticed this phenomenon on both my Helios and Pultec EQ designs.

The EQ uses a series RLC  circuit to provide boost. Below resonance the C becomes open circuit and above it the inductor becomes high impedance. HJowever, there is usually a parasitic C across the inductor winding which forms a parallel resonance at some high frequency. This is the self resonant frequency of the inductor and should normally be well above the audio band. However, I have found that with some relatively high value inductors from Carnhill this resonance can be in the 20 to 30KHz region.  Theinital effect was to apparently shift the respected resonance of high frequency boosts such that the series C had to be reduced in order to reach the required frequency. It was not until much later when someone made measurements up to and beyond 48KHz that this effect came to light.

Cheers

Ian

analogguru said:

What ? 18kHz in the "mid-band" ? 

Click to expand...

Actually, 18kHz is in the midband between 9K and 25k, which are the designated center frequencies.

analogguru said:

BTW, I bet that the secret "AIR BAND ®" boost is nothing else than something like an aphex exciter....

Click to expand...

And you would loose your bet. The so-called Air band is a shelving EQ with restricted boost, which results in very gentle slope, when the Aphex Aural Exciter uses a 2nd-order high-pass filter that drives a "harmonic generator" (in other words a glorified fuzz box).

abbey road d enfer said:

Actually, 18kHz is in the midband between 9K and 25k, which are the designated center frequencies.

Click to expand...

Cool, nice to learn everyday something new.

abbey road d enfer said:

And you would loose yout bet. The so-called Air band is a shelving EQ with restricted boost, which results in very gentle slope, when the Aphex Aural Exciter uses a 2nd-order high-pass filter that drives a "harmonic generator" (in other words a glorified fuzz box).

Click to expand...

Interesting, I was conviced that I wrote "....else than something like an aphex....", must have been lost somewhere.....

Aha, but the Aphex B fuzz-box-harmonics-generator was not used anymore on later exciters and not on the mentioned clones like Dualfex, Ultrafex from Behringer or the EXR EX3 or......

paranoidandroid said:

This is the original that Peter posted a long time ago which to me is very difficult to read, but maybe your eyes are better than mine!

Aha and here is from the publicly available users manual. BA 211 is midband. You can see which sections he cut out, aside from the mic amp, BA194 the LPF section:

Click to expand...

From this, I can see High and Low-pass filters, a Bass (LF) control and a Mid control. How is this supposed to allow 25kHz EQ?
It seems like each band is handled by a separate circuit, so I would think that rules out interaction; the plot thickens...
EDIT: after putting my glasses on, I found that the HF shelf shares the same circuit than the LF shelf (as is typical in a Baxendall-type arrangement). So, the HF shelf doesn't share anything with the MF section, thus no interaction can happen, other than the expected algebraic addition of amounts of B/C.
Since the notch shows on the graphs of MF alone, it leaves no doubt that it's due to a parasitic in the MF stage. Indeed, Ian's explanation is the most plausible.
I would think one could probably live with that quirk and never realize there's something not quite right. OTOH you seem to have been bothered by it.

ruffrecords said:

This could be due to parasitic capacitance in the EQ inductor. I noticed this phenomenon on both my Helios and Pultec EQ designs.

Click to expand...

Thanks Ian for this insight. I was wondering if this were the case, if it would then make sense to move the high shelf section to the end of the signal chain instead of the beginning, in order to avoid the mid band inductor capacitance interacting with the boosted highs. But then of course you have the issue of boosting all the noise created by the other bands in this case?