Basics? Is there a preferred order in which the various analog filters are switched in Pre-Mastering process?

[farmersplow]

As I am new to the forum, I would like to introduce myself briefly.

I come from Austria (Vienna) and am 50+ years old; and vinyl fan for almost as long.
( My English is not very good, so I beg your indulgence).

During the corona lockdown I had some time and started to work on the development of a record cutting machine (someone who tried again to develop a lathe). According to the motto: "The farmer and his plow". I would also like to create the necessary framework conditions (plate heating, pre-mastering filter, suction, feed control, signal testing, filler font method etc.).

During my research I came across this extremely interesting forum - there was no getting around it, because obviously the concentrated load of know-how in the field of record cutting is located here.

Since I have gained some experience with the development of audio filters, but have little experience with their use, I wanted to put my question to the forum right away:

Is there a preferred order in which the various analog filters are switched?

For my pre-mastering, I have different filters for signal processing before the processed signal is sent to the cutter head. But I have no experience in which order the filters should (ideally) be set.

The following filters:
1. Inverse RIAA (developed by Wayne Kirkwood)
2. Elliptic Equalizer (KA Electronics EEQ-12)
3. High-Pass (Low cut) (self made -24dB 4th Order; 20, 30, 40, 60, 80Hz)
4. Low-Pass (High cut) (self made -24dB 4th Order; 9, 12, 15, 18kHz)
5. Limiter (self made; in progress)
6. Power amplifier

Is the specified order ideal, or does it matter in which order the filters are set (except for the limiter and power amplifier, which must always be at the end before the signal is sent to the cutter head).

I would be very grateful for any advice.

Thomas from Vienna.

[dmills]

It matters from a noise and headroom perspective as well as an 'what exactly are you limiting' one.

Generally the elliptical eq goes in first, just because why not? This is really part of the transfer console for me, not the lathe.
Then I would typically high pass, apart from anything else, this will strip off any DC or subsonics that do nothing except cost headroom later.
For me I then lowpass, do the main bandwidth definition before the dynamics (Actually ALL of my processing cards have 100KHz second order LPFs at the input to limit the HF response, and the LPF card has high order inverse Chebyshev going over at 50kHz as the ultimate limit).

In my design I then limit (Really de-ess) with a three band deesser with limits that fall at 6 and 12dB/octave to limit velocity and acceleration to the levels imposed by the geometry.
Then IRIAA, Then Feedback error amp (work in progress, and needs to match head design), then power amps, driver protection and off we go.

In a non feedback chain or one lacking a properly configured de-esser I might slap the limiter between the eq and the power amps, but you CANNOT really put dynamics inside the loop in a feedback system, nothing good comes of it.

I would note that simple non feedback machines typically wind up needing something a LONG way from a classical IRIAA curve to work right, but you sort of need to measure and then design a corrective EQ for those, generally it looks like what would happen if you dropped the IRIAA by 6dB/Octave and scooped out a honking great notch where the mechanical resonance is.
For a feedback head, the output of the IRIAA network is a velocity signal,

[farmersplow]

Thank you for the quick and very helpful answer. So far your explanations are all clear to me.
My filters are 4th order Butterworth which (hopefully) also have a steep curve. For the time being, I have not planned real de-essers (because I hope to get by with the low-pass filters).
In my first project phase I will still be working without feedback. So I would be very interested in what you mean by "something LONG way from classical IRIAA curfe"? In any case, I still have a lot of measurement work and adaptation work to do.
In any case, thanks for the answer.

Thomas

[dmills]

Feedback heads are typically mechanically very under damped with much of the damping being due to the error amplifier circuit (In pursuit of more sensitivity at high frequency).

Non feedback heads are typically heavily mechanically damped to try to take out the worst of the generally massive mechanical resonance at a around a kHz or so. Below resonance they are basically stiffness controlled, and so have a flat groove amplitude Vs frequency response (Velocity rises at 6dB/octave), above resonance they are mass controlled and so have a falling amplitude response (flat velocity response) until they hit the mess of high frequency resonances, anti resonances and general dogs dinner up above about 5k or so.

First order it is an IRIAA without the low frequency slope, more or less, but that is a very idealised view of the thing.

What this means in practice is that cutting some very low level pink noise (or a sweep) and then loading the result into your DAW and doing an fft will be necessary before you get too far into figuring out the eq board design, particularly if you are doing a non feedback version of a design that is normally a feedback head.

[farmersplow]

I understand. I hope to have enough mechanical damping in the head to reduce resonances. Also the volume and frequency of the pink noise I will see (hear) and analyse when the head is finished. Your informations and details are very helpfull for me and I want to thank you for it. When I have the first results, I will post them here.

[markrob]

I like to use program material rather than noise. At high frequencies, its easy to mistake the system noise floor for the actual signal. This can result in incorrect EQ. If you use a noise source, make sure you establish the noise floor first via a silent cut. Then make sure your signal is 5-10 db above the floor to insure you are seeing the true response.

[farmersplow]

It was my plan to measure the noise floor (without signal). I think that there should be as little background noise as possible here. The corresponding filters must be matched to this - I understand that. The information that the signal should then be at least 5-10 dB higher is very helpful. - Thanks!
As soon as I have the first data, I'll post them.

[farmersplow]

have changed my mind and will now work with a feedback coil after all. I will present my progress in another post. - Thanks again for the suggestions.