Pitch Control

[Dub Studio]

Quick survey, what signal are you guys sending to the pitch controller?

I am experimenting with sending the signal to the controller, then delaying the signal one turntable revolution before cutting.

Anyone doing it differently to this? If so, what is the rationale behind it?

I am particularly interested to hear if anyone is using a different timing to this, or a different signal.

[opcode66]

Are you developing your own P/D computer? It is not easy, but certainly possible...

You send equal volume signals to both the cutterhead and the p/d comp.

Various pitch systems have used different timings. There is no ideal timing here. It really just depends on how you are going to be accomplishing this and to what degree of accuracy.

The Neumann system splits one rotation up into quaters. It looks ahread 3 quarters as it cuts one quarter. So, the delay timings are 1.1 seconds at 33.3 and 0.84 at 45.

Essentially you need to determine a few things about the audio within those partial rotations. Things like intensity of the audio (especially of the lower frequencies as they tend to eat up the most lateral space on the disc at high volume). Also, you need to know how much stereo separation is going on at any given time. That tells you how deep the stylus is since out of phase audio causes the stylus to push in or pull out and therefore affects the groove width and thereby can affect the lateral space required for a groove. Also, you might want to limit how much the stylus can pull up off the disc in order to maintain an minimum cutting depth.

[Dub Studio]

hehe thanks for the info - I have no programming skills so I think a computer is probably beyond my capabilities, but I am hoping to use some simple logic and a little ingenuity to improve on simply sending the signal to the pitch controller one turn ahead of the recording path.

This system is obviously flawed because it cannot sense how the groove being cut will interact with a) the preceding groove, and b) the following groove - all it can do is move the head on sufficiently to give the space that one isolated groove would theoretically need if the preceding groove, and the following groove were unmodulated. Is that correct?

I guess what I am looking for is information on how a computer calculates how much space to leave so that the preceding groove doesn't clash with the current groove, while at the same time ensuring enough space is left for the following groove.

[Serif]

...I guess what I am looking for is information on how a computer calculates how much space to leave so that the preceding groove doesn't clash with the current groove, while at the same time ensuring enough space is left for the following groove.

Fwiw, the LS-76, like a few others, requires that the advance copy of the signal to be cut be received by its feed and depth computer early by one turntable revolution (e.g., 1.8 seconds, for 33-1/3 rpm). For 30 ips tape, this would require a 54" spacing in the threaded path between the advance repro head and the modulation repro head (which is ultimately sent to the cutting amps).

In addition to the revolutionary arrival time of the advance copies of the signal, there is also required a copy of the Left channel Program signal (the same signal as the Left advance signal, only arriving at the same time as that which arrives at the cutting amp). This Left Program signal is used in real time to keep track of the inside wall's lateral component. I guess memory was expensive back in 1976. 2 seconds was the max for the digital audio converters used (10 kHz sampling rate! 8 bit words!).

The inside wall of the groove has nothing to look out for, since there should be only virgin wax between it and the spindle. However, the outside wall of the groove has to be cut in such a way that it doesn't twine with or overcut into the already cut groove (one revolution behind the "live" action).

One _also_ needs a copy of the Left Advance signal (timing-wise) in order to do a L/R difference calculation in order to determine the out-of-phase content so that the depth changes required to preclude both lift out and "pearling," or even precluding cutting more deeply than is adequate (Helas, "deeply" cut grooves are unwanted, since they waste time and/or fidelity). The LS-76 contains an ITC audio transformer in the back of its card cage for the purpose of deriving the Stereo (Difference) signal from the Advance inputs...

The LS-76, therefore, has four separate feed and depth computing inputs: Left Advance; Left Program; Right Advance; Right Advance. Two copies of Right Advance are used to allow independent adjustment (with extra attenuators) of the feed and depth signals to be "offset" from that which is their calibrated set points... One may calibrate the system for a Base Feed and Base Depth which can be modified for unique cutting jobs, where program requires, without having to recalibrate the whole system.


Compudisk (by parametric eq co-designer/ co-inventor, Burgess Macneal) and Zuma also use the full-revolution (1/1) look ahead technique. Unless memory is applied, I don't know how else one can preclude overcuts or twining, on the one hand, and enable stingy groove-packing - even "spooning" - of in-phase sustained notes (such as those from a church organ), on the other, unless the full revolution is contemplated.


- Andrew

[dubcutter89]

Hey!
I had a qick look at your website - do you use souris auto-pitch or anything else?

As it has been already said: There are a lot of auto-pitch systems and most of them have their own way of accomplishing "correct" spacing

A simple system may only rectify a signal and do some low pass filtering on it...
A more complex may split the turntable in pieces and store the max required space in each segment...
ASFAIK state of the art systems (Neumann VMS80+, Vinylium Pitch) even analyse the phase relation between each groove so they could even be stacked within each other...
Polygram also did calculations on how grooves can overcut without causing skipping!


And about timing:
I know systems that don't need preview (asfaik the kingston cutter has a very small auto pitch incuded...)
Systems with a preview of only ~0.1 seconds
The "stndard" ones with 0.5-1 tt revolution
and software that calculates ideal pitch for a whole side of the disc, and therfore needing the whole context of a side prior to cutting...


about basic calculations:
we know that the left groove wall is corresponding with the left channel
and the right one with the right channel.

cutting from right to left you want to know the left-channel of the previous groove (L-) and the right that you actually cut (R+).

So the minimum needed distance is D = (L-) + (R+)

...and now add some offset, vertical modulation, preview, electronics and there you have a auto pitch computing system

Lukas

[Serif]

...
ASFAIK state of the art systems (Neumann VMS80+, Vinylium Pitch) even analyse the phase relation between each groove so they could even be stacked within each other...

The LS-76, Compudisk, and Zuma II are also sensitive to "absolute polarity".

And about timing:
I know systems that don't need preview (asfaik the kingston cutter has a very small auto pitch incuded...)
Systems with a preview of only ~0.1 seconds
...

Memory is much longer, now. But I wonder how these myopic lathes can ramp sufficiently without audible artifacts and with efficiency.

about basic calculations:
we know that the left groove wall is corresponding with the left channel
and the right one with the right channel...

From the context of the position of the cutting head, yes. However, a "left groove wall" doesn't quite exist. There is an inner groove wall and an outer groove wall, however, which remain fixed no matter how you "lens" the disc. When the cutting head is dropped onto the blank, the left channel audio is engraved and impressed into the left-hand wall, but if you look at the disc from the other side of the platter, where the pickup (tone arm) is, the left channel is, there, in the right-hand groove wall. Round and round it goes, but the left channel is always in the inner groove wall when the standard disco-choreography is observed. It might have been Left, "in," but at least nothing was Left, "out." (;

cutting from right to left you want to know the left-channel of the previous groove (L-) and the right that you actually cut (R+)...

I agree if you are saying you have to know the left channel history and the right channel future. Not sure if this is what you are describing, above.

So the minimum needed distance is D = (L-) + (R+)...

The Difference between the left and right is normally cited as the vertical component of the stereo signal. This affects feed requirements due to the large stylus included angle.

If I had to do the algebra, I'd say feeD(min) = L(past peak) + R (present peak). The only way to "get ready" for the upcoming right channel's present requirements is to analyze it ahead of time. The left channel behavior will only be measured until the following revolution at which time it will be acted upon as its data impacts the next groove of the right channel modulation (which are the stylus excursions that move away from the spindle).

Since we need to know at minimum one full revolution's information or else possibly waste space (since we dare not assume we'll need less than we might) I don't yet understand the way the short-time computers are handling this...



- Andrew

[dubcutter89]

well, thanks for clearing up my post...

about those short timers:

theres one paper i have about a very early design (published 1954) which says 0,1 sec preview. very interesting full analog design with tape loop storage and memory up to 10 revolutions! Don't ask me if those numbers are right, or if it works well...

the no preview solution, i guess, is something like fixed pitch with automated decrease in pitch if the previous groove allows for it...

Lukas

[Dub Studio]

Hey!
I had a qick look at your website - do you use souris auto-pitch or anything else?

Thanks for all the great info! Yes I am using a vinyl recorder, and after nine years of cutting with fixed pitch, I thought it was about time I got some pitch control

[Dub Studio]

OK I have been tinkering around this afternoon with the aim of getting a nice DAW-based pitch "computer" (I use this term loosely as I am no programmer and I am relying on nothing but the audio signal).

I thought the best way would be to devise a system that not only looks ahead, but also looks backwards, thereby dividing the task of pitch control between what's coming up, and what's gone before.

The aims are to control my variable pitch unit:

- to be able to go from really loud to really quiet without any kissing / twining
- to be able to go from really quiet to really loud without any kissing / twining
- to nest any consecutive lateral modulations wherever possible
- to nest consecutive vertical modulations wherever possible
- and of course… to save land wherever possible!

I am pleased to say I devised a quick and simple system to achieve all these things, and it works pretty well, here is a picture of test cut:

http://instagram.com/p/S1GNbAwSZL/

From left to right it shows a cut of a sine wave of roughly 500Hz @ 33rpm, which of course is almost perfectly in phase, so its nicely nested. Then I switch the polarity of one channel so now its not in phase, and the grooves spread out accordingly. Next I up the frequency to approximately 513Hz, and the resulting vertical modulation now nests nicely too. Then finally I switch the polarity again, and the grooves spread out again to accommodate the resulting lateral modulation.

All in all I am pretty pleased with the results, although it definitely needs a fair bit of tweaking. I have just cut some pretty wild dustup and it didn't like some of the transients!

[Virtalahde]

Just a thought from this morning:

I don't remember anyone doing a "pitch sequencer" - a simple unit that would hold a number of different LPI settings in a row, each dialed on a separate pot. By pressing a button, the sequence would advance to the next LPI setting, which could mean a passage in a part of music or a different track, whatever the engineer has specified.

Such a sequencer is easy to build and hook up to a lead screw motor controller, and it would give much more repeatable settings for cutting the same disc a number of times. If you want to go fancy, the next sequence button could easily be automated with a trigger signal, which could come from your DAW. At simplest it would be a short click of sound coming from a different output, converted to a trigger signal with a simple box.

Yep, I come from the analog synthesizer world.

[Serif]

theres one paper i have about a very early design (published 1954) which says 0,1 sec preview. very interesting full analog design with tape loop storage and memory up to 10 revolutions! Don't ask me if those numbers are right, or if it works well...

Echoplex-plex-plex-plex-ple....

the no preview solution, i guess, is something like fixed pitch with automated decrease in pitch if the previous groove allows for it...

That's what some of the standard fancy programmers can do - Contaction. I think this is VMS80 spec? TubeFan might be able to chime in, here? They have a Base Pitch, which can be goosed, for normal expansion requirements, but, in case there's a passage of music which is sufficiently "piano" (rather than "forte"), a de-expansion, below the Base Pitch, as in, contraction, is possible, it will do that. Formid...


- Andrew

[gold]

a de-expansion, below the Base Pitch, is possible, it will do that.

Base pitch means that the pitch won't go below that. That's the point. A VMS80 doesn't work from base pitch like a VMS70. You set "minimum land" for groove spacing.

[opcode66]

You achieved what you showed in that picture in one afternoon with strictly analog circuitry???

[Serif]

a de-expansion, below the Base Pitch, is possible, it will do that.

Base pitch means that the pitch won't go below that. That's the point. A VMS80 doesn't work from base pitch like a VMS70. You set "minimum land" for groove spacing.

Hi SALT,

The way we are using the terms, "...minimum land" should also mean that the pitch won't go below that, technically speaking. If it only means "always seek the minimus," then it is not doing "contracting," or "de-expanding below" a base pitch (as in, seeking a higher count of lines than normal), and is, therefore, not germane. It might be German, however. o:

I prefer saying Feed, but I don't see Base Feed written down, even in the manual which calls Pitch, Feed.

I haven't seen any sub-basements, yet. But I know that I want one.


I have seen basements with sub-minimal masonry. }:


-Andrew

[Dub Studio]

You achieved what you showed in that picture in one afternoon with strictly analog circuitry???

Not strictly no, I had been working on the general idea for a few days, but decided to try and nest the vertical modulations that afternoon.

I am working from a DAW so I was not using analogue circuitry, but I guess the same system could be made with analogue if you could have a 3.6-second analogue delay.