OK, but is the cutter head linear? What do we know about how the groove is cut?
Was reading some stuff about various 4-channel LPs of the 1970's. Apparently JVC developed ½ speed mastering and "special" vinyl for their CD-4 format. Frequency response of JVCs format extended to 50 KHz (the rear channel information was encoded above 20 KHz).
The ½ speed mastering later evolved into the first 2-ch "audiophile" recordings at MFSL (they did put out a very few 4-ch titles).
Not sure what that means WRT this project, just throwing it out there.
See:
Quadraphonic Systems
The ½ speed mastering later evolved into the first 2-ch "audiophile" recordings at MFSL (they did put out a very few 4-ch titles).
Not sure what that means WRT this project, just throwing it out there.
See:
Quadraphonic Systems
Last edited:
It is very affordable to get a single LP cut at one of the dub plate manufacturers.
It's about 80-100 Euros for one LP in highest quality on a flat PVC blank. The PVC has the same long life time as regular vinyl.
So everyone that want's his own test record can have it cut and no pressings (with additional process variations) are needed at all.
This is a very good test to see how your plans work out with only a minimum of costs involved.
However, as my previous input already didn't shake a feather here I'll leave the further discussions to you guys
It's about 80-100 Euros for one LP in highest quality on a flat PVC blank. The PVC has the same long life time as regular vinyl.
So everyone that want's his own test record can have it cut and no pressings (with additional process variations) are needed at all.
This is a very good test to see how your plans work out with only a minimum of costs involved.
However, as my previous input already didn't shake a feather here I'll leave the further discussions to you guys
The problem is no matter what measurement method is used, real playback swept f response (and distortion) can depend upon test modulation level - and a proper test should be able to report this. To do this, the test record must include swept level of swept f.
For sweeps, perhaps 6 or 10 seperate overall levels spanning the dynamic range ?
Spectral noise test would be far easier to sweep in level, allowing a 2D waterfall like display....... with all the pitfalls of spectral noise tests though. Opportunity to address some of those pitfalls too though.
The seperation of stylus motion into lateral and horizontal vectors is entirely artificial and for analytical convenience. In reality, stylus motion is free in 2D, centred around some nominal radial centre. All vector angles are possible.
However, not all vectors are equal because the groove wall can't exert downforce on the stylus - only downforce is VTF applied via the cartridge suspension spring/damper. Trackability is worst for pure vertical stylus motion then, and (depending on level) there's no guarantee that a sweep will be trackable at all frequencies.
Also, vertical motion of the cutterhead is not normal to the plane of the record, for cutting reasons. Generally offset by about 20 deg, but AFAIK lacquers are said to have have some spring and variability. The result of this error, plus physical interaction of stylus front/rear bulk, means distortion for pure vertical motion is pretty poor and increases significantly with vertical curvature (ie frequency and level).
For all of these reasons, vertical groove shape is 'managed' during cutting, especially for large excursions, and in real programme material there's not much out-of phase L-R content anyways. Fortunately. Thus it all works and sounds fine in practice. But measurements would reveal horrors, I suspect, unless representative of true programme material and mastering restrictions. Thus ~300Hz, perhaps, for a test tone.
Sometimes test tracks around ~300Hz are also a trackability test, a means of characterising how much VTF is required to keep stylus in vertical contact with grooves.
Most test records seem pretty fine, IME. Presumably the big houses such as Shure, Decca, NEC etc etc had influence and control of the mastering/production process for test records, and able to overide normal constraints if appropriate. A 'people' test record as we are discussing might need access to similar overides, methinks. I seriously doubt that rocking up with a set of PCM test files would result in an accurate transfer without it. Rules are there exactly to stop people like us..........
LD
Last edited:
Fairy nuff. This has got me thinking about a rolling modulation. You could (up to a certain level) rotate a 1KHz modulation through 180 degrees over a period of a couple of revolutions to look at how the wiggly bit handles the different vectors in 2D space. But if it doesn't help us in terms of coming up with the ideal cartridge beyond that which we can find out from existing tests doesn't help...
Oooo yes, I like that idea........ the rolling vector.......
Could well also be other tests that idea applies to....
LD
It is, but as you mentioned in an earlier post, we need a good relation with the mastering engineer. We need someone who understands that we need precise things, not simply "optimized."
And I agree, we may have to pay for a few revisions of the master before we get what we want. As long as we pay for it, there shouldn't be any problems with the cutting facility.
I like the idea of sweeps at various levels as that would be useful to me. Glad it has come up in discussion.
What would be their amplitude and duration? Is 5.4 seconds (3 revolutions) long enough? That works for me.
And amplitude - something like +6, 0, -6, -12, -18, -24 dB taking 0 as 8cm/sec or whatever is chosen as reference on the LP.
What would be their amplitude and duration? Is 5.4 seconds (3 revolutions) long enough? That works for me.
And amplitude - something like +6, 0, -6, -12, -18, -24 dB taking 0 as 8cm/sec or whatever is chosen as reference on the LP.
Yes maybe, or in a little bit. I've done some reading over there to try to gain some understanding of the challenges
From what I've read, a 1kHz signal cut at 5cm/sec has an excursion of +/-11 microns. How does that fit with what you can cut?
I was trying to work out how to synthesise a stereo track for this over 360 deg..........so that the vinyl vector of stylus motion full rotates once every few seconds?
LD
Are you sure this can be done without PM? Two exact 1kHz sines have a single phase relationship vs. time.
Ah, thanks Bill I see why you said 180 deg now, because for any symmetrical waveshape, eg sine, the motion would be duplicated and so 180 deg of vectors involves motion that covers the full 360 deg 2D stylus motion range.
So what one has to do is progressively pan stereo width from mono to pure antiphase, to produce the effect.
Then the phase of L&R channels needs to slip one cycle over whatever time is chosen for the rotation. Since frequency is rate of change of phase, one channel needs to be slightly off the frequency of the other to produce this effect, by one cycle per unit time of the rotation.
At 1kHz, and a rotation of 1s, one channel needs to be synthed at 999.5Hz and the other at 1000.5Hz - I think...........?
LD
So what one has to do is progressively pan stereo width from mono to pure antiphase, to produce the effect.
Then the phase of L&R channels needs to slip one cycle over whatever time is chosen for the rotation. Since frequency is rate of change of phase, one channel needs to be slightly off the frequency of the other to produce this effect, by one cycle per unit time of the rotation.
At 1kHz, and a rotation of 1s, one channel needs to be synthed at 999.5Hz and the other at 1000.5Hz - I think...........?
LD
You are ahead of me. I'd got as far as moving from mono to antiphase, but had missed that, to swing over the top and back to mono is 180 degrees of rotation if you think about the summed output, but 360 degrees of phase rotation. But I'm stuck visualising the two points which are single channel modulation occuring at 90 degrees phase rotation.
These give the linear combinations of vertical , lateral, L, and R. Maybe just adding L and R in quadrature would add some information? If this was a perfectly linear system there is too much information gathered, but if not what captures the most useful information.