"Corrective EQ"

[RuudNL]

Would you like to point out a 'real life' case where this is a disadvantage?

In case the amplifier input in unbalanced. And is amplifying the 'wrong' (FET source) signal.

Why don't you make something with my method first?  It's easy.  Then you have a baseline to see if you can do better.

For low frequencies, the FET would then act as a phase splitter, but for high frequencies the FET would act as a source follower:
This works. Disadvantage is that the attenuation of high frequencies only works on one 'leg' and that the maximum attenuation can be no more than 6 dB.

I already confirmed the solution you suggested works.

 

[ricardo]

Would you like to point out a 'real life' case where this is a disadvantage?

In case the amplifier input in unbalanced. And is amplifying the 'wrong' (FET source) signal.

If your preamp is unbalanced, it will complain & give yucky results on loadsa other microphones too so this is the least of your worries.

Got any P48v preamps which are unbalanced?

 

[Whoops]

Nice thread, thank you all for the contribution and info.
It's a shame the discussion didn't continued.

 

[rogs]

There seems to be 2 different approaches  to adding corrective EQ internally to mics...

One is where an attempt is made to copy the style of circuit applied to some of the 'classic' vintage mics.
That is normally some kind of frequency selective negative feedback, and of course is really only appropriate when it is applied to the same type of capsule as it was originally intended for.

The second is where passive HF attenuation is applied to compensate for the overly 'bright' response from many cheap Chinese K.67 style capsules. Those capsules seem to vary enormously, so specific component values will often be capsule dependent.
Henry Spragens made some interesting notes on EQ which you can find here: http://www.audioimprov.com/AudioImprov/Mics/Entries/2014/2/7_Mic_Electronic_Eq_Pt.1.html    (for
subsequent pages click the NEXT box on each page).
Unfortunately - as has already been pointed out by Ricardo and others - the introduction of series resistors into the signal path will degrade the noise. Henry suggests this is not really a problem, but I personally found find that using anything higher than around 1K will introduce significant noise.
And of course it is only a simple HF attenuation technique which - as Ruud has already mentioned - will often result in loss of extreme HF content, as well as the mid range boost you are trying to control.

I have to say, I am a bit surprised this approach is still so widely discussed.  Much better, I would suggest, to introduce EQ externally?
That could be in post in your DAW when the mic has been recorded.  The comprehensive EQ  offered by most DAWs should allow for a complex correction curve, if required.

A second option - for live performance for example - would be to introduce hardware EQ into the signal path. Most modern mic channel strips or audio interfaces include insert points,  which will allow the introduction of EQ  - as simple or complex as required - into the signal path at line level.

Taking that approach allows for much more flexibility and removes any problems that adding internal EQ can cause from noise or balance issues within the mic itself.

I have built a number of mics using my own variation of the Schoeps circuit, and had added an internal passive LPF in the early versions. Sadly that did introduce some extra noise, so I now make 'linear' response circuits, and apply external EQ to suit each individual capsule type, as required....either in post, or in hardware.

 

[RuudNL]

Got any P48v preamps which are unbalanced?

Yes, recently I had a preamp made by 'Analog Addicts' here for repair and they did the following:
Two 6.8 K resistors to +48 V, one input leg via a 47 mfd capacitor to ground, the other input leg via a 47 mfd capacitor to the (tube) preamp circuit...  :

(By the way: the whole construction looked like a pile of dog sh*t!)

 

[Khron]

One must wonder what sort of "magic smoke" they must've been on... 

Yes, recently I has a preamp made by 'Analog Addicts' here for repair and they did the following:
Two 6.8 K resistors to +48 V, one input leg via a 47 mfd capacitor to ground, the other input leg via a 47 mfdcapacitor to the (tube) preamp circuit...  :

(By the way: the whole construction looked like a pile of dog sh*t!)

 

[abbey road d enfer]

Yes, recently I had a preamp made by 'Analog Addicts' here for repair and they did the following:
Two 6.8 K resistors to +48 V, one input leg via a 47 mfd capacitor to ground, the other input leg via a 47 mfd capacitor to the (tube) preamp circuit...  :

Well, it's not balanced, but it should work somewhat. Anyway, I wouldn't pull a hair over caring for such a misguided design.

 

[abbey road d enfer]

I totally suscribe to the idea that external EQ, either in hardware or software is the best way of dealing with the HF resonance of condenser mics, because it's a 2nd-order resonance, that no 1st-order EQ can really compensate. OTOH, I understand that some users want to have the flattest possible response, although the motives are often a tad dubious. The most common argument is that "an acoustic problem cannot be compensated electronically", or "I want the shortest path from the air to my ears".
Whatever the motives, I believe it can be done passively by using well-defined RLC circuits, which could not be installed in a SDC, but could find place in most SDC's.
See attached circuit.
This would work with a commonly available 10mm inductor
https://www.mouser.fr/Passive-Components/Inductors-Chokes-Coils/Fixed-Inductors/_/N-wpcz?P=1z0wpwc

 

[kingkorg]

Something similar is done in Line Audio CM4 to achieve linear response, and it works like magic.

 

[analogguru]

I used the combination of 23(22)mH and 18nF in a slightly different arrangment simply because  adjustable 23mH coils can be found in the 19kHz-filters of old Dolby-B units....

 

[Whoops]

I totally suscribe to the idea that external EQ, either in hardware or software is the best way of dealing with the HF resonance of condenser mics, because it's a 2nd-order resonance, that no 1st-order EQ can really compensate. OTOH, I understand that some users want to have the flattest possible response, although the motives are often a tad dubious. The most common argument is that "an acoustic problem cannot be compensated electronically", or "I want the shortest path from the air to my ears".
Whatever the motives, I believe it can be done passively by using well-defined RLC circuits, which could not be installed in a SDC, but could find place in most SDC's.
See attached circuit.
This would work with a commonly available 10mm inductor
https://www.mouser.fr/Passive-Components/Inductors-Chokes-Coils/Fixed-Inductors/_/N-wpcz?P=1z0wpwc

Thanks Abbey,
that actually seems more like a proper correction EQ for the K67 capsules high boost peak.

I also understand what Ruud expressed, that a Low Pass filter, will tame the presence but will also cut a the high end above the Presence Peak, or above the problem area.

Could you please just explain on how in that network could the Frequency and the cut be changed or tuned?

Thank you so much

 

[mhelin]

In theory the noise in microphone is caused (mainly) by the diaphragm damping resistance. In K67 type capsules it's lower than with other capsule types so it will cause undamped peak at resonance frequency. This can be corrected using a notch filter with a proper frequency and Q - the problem is to find them and then calculate the correct component values as the capsules vary so much. How could this be done in practice? Maybe if you can measure the FR in the resonance area and insert an external EQ with variable frequency and Q it could be done and maybe automated. B&K and other use even less damped capsules than these Chinese capsules are and corrective EQ to make almost 0 dB (A) noise level microphones (mostly for some measurement purposes).

 

[abbey road d enfer]

Could you please just explain on how in that network could the Frequency and the cut be changed or tuned?

Frequency is governed by the formula F=1/2.pi.sqrtLC
For one frequency there is an infinity of solutions, so one must be chosen that gives the correct BW (notice I don't use "Q").
For example halving the inductor to 50mH and doubling the cap to 7.8n would about double the BW.
The amplitude is governed by the total DC resistance, i.e. the sum of the resistor and the DCr of the inductor; the higher the resistance, the lower the cut. E.g. increasing teh resistor from 2.2k to 4.4k reduces the cut from 5 to 3dB.

 

[abbey road d enfer]

This can be corrected using a notch filter with a proper frequency and Q - the problem is to find them and then calculate the correct component values as the capsules vary so much.
How could this be done in practice?

In the U67, tuning is done with one single component, capacitor C17. I tend to conclude that Neumann have a way to reduce the tolerance on their capsule's tuning in such a way that ultimate tuning is limited to that single component.

B&K and other use even less damped capsules than these Chinese capsules are and corrective EQ to make almost 0 dB (A) noise level microphones (mostly for some measurement purposes).

Some have used these mics for music recording purpose and reported disappointing results. Since frequency response cannot be suspected, it seems that transient response (waterfalls) are the culprit...

 

[kingkorg]

The original vintage Neumann's k67 does not have a symmetrical peak at high end. It gradually rises towards 10k, and then drops much sharper.

When you use  typical de emphasis filter, you end up with pretty flat response up until 13k. And i am not talking about Ai version.

That peak can not be addressed with notch filter. However, many chinese capsule can.

One needs to  have pretty reliable gear in order to measure the true capsule response, and then design a precise filter for it. Even grille can cause some peaks that show in measurements if things are bit off. However in real world situation, they are not there.

I have experimented with some notch filters, but found that different FETs respond different with the same components in the circuit. Also that you can not really count on simulation software here when it comes to Q and precise frequency.

Some Schoeps based circuits like Behringer B2, are slightly different and have actually a BJT in place of FET. They also respond differently.

 

[abbey road d enfer]

The original vintage Neumann's k67 does not have a symmetrical peak at high end. It gradually rises towards 10k, and then drops much sharper.

When you use  typical de emphasis filter, you end up with pretty flat response up until 13k. And i am not talking about Ai version.

That peak can not be addressed with notch filter. However, many chinese capsule can.

Actually, a notch filter alone cannot address the whole issue, but neither a LPF alone. Neumann's choice is a compromise (at the detriment of the top octave) that fixes most of the issue with a circuit that is not too complex, and easy to tune.

I have experimented with some notch filters, but found that different FETs respond different with the same components in the circuit.

Indeed. The performance of a typical single FET-based active notch filter depends on the FET's gain and input capacitance.

Also that you can not really count on simulation software here when it comes to Q and precise frequency.

The results of a simulation cannot be better than the models it uses. Some FET models are way off, and anyway, manufacturing tolerances on FET's are typically -50/+100%.

 

[Recording Engineer]

All interesting, from a user perspective as well. While I generally agree a somewhat balanced response is generally desirable, I am always actively looking for very-EQed mics as well for routinely-problematic applications and kind of eq with the mic what I’m normally trying to do outside the mic. For example, some of the Nevaton mics have pretty-wild responses I’d love to try!

And while not for a problematic application at all, my new Nordic Audio Labs NU-100k is pretty amazing with its EQed response!

 

[mhelin]

And while not for a problematic application at all, my new Nordic Audio Labs NU-100k is pretty amazing with its EQed response!

Do you mean it's response EQ'd? Actually Martin told in a FB video that they developed a new capsule to get rid of the resonances and all pre-EQ's stuff (compared the NU100K to U47 at that point and claimed that the resonance of U47 cannot be EQed out) so that it could EQed in post. Anyway, would like to know more about the capsule.

What comes to that if a resonance can or cannot be EQed out depends on the resonance, is it minimum phase or not.

 

[kingkorg]

What you see here is c800g FR. Notice that 180° response is just about 10db down at 7K-10K compared to 0°.

Keep in mind that is in anechoic chamber. Now imagine in less ideal environment, possibly with a glass window behind mic, and source being further away does not help either. I can imagine that peak being somewhere between -3 and -6db in real world.

This is also a pretty good capsule, i can think of many worse.

You can try to compensate for that with an EQ, but compared to response at 1K where rejection is excellent, there will always be some 7-10K weirdness. 

 

[Whoops]

8K weirdness.

8K weirdness, 8K Peaks, 8K boost

Are all stuff I really don't want when recording vocals