Peak velocity for one channel only is (peak displacement*2*pi*frequency).
11.2e-6*6.28*1000=7cm/sec peak velocity.
For mono this velocity this reduces to 7/sqrt(2)= 5cm/sec.
So 45u pp or 22.5 u peak, gives 22.5e-6*6.28*1000=14.14cm/sec for a L or R only signal.
In Mono this would become 10 cm/sec.
Hans
I agree, that isnear the practical limit of playback tracking performance. Limiting factor is the peak angle the groove makes with the tangent, in this case ~ 50 deg depending on where the track is located on the record.
In the candidate track list, it's covered in the 'groove angle' section at the high end ~50 deg. It will work out to be effectively the same track as the CBS test record.
The point of having multiple tracks with graduating groove angles on the test record is so that one can determine cart/arm performance in terms that are applicable to groove angle generally. And instantaneous groove angle = instantaneous programme level in real programme material. One aspect of trackability performance, and tracing error.
I think angle is a better way of looking at the same thing: it's very human to think in terms of peak groove angle, and it's frequency independent so easy to apply.
LD
Pano, you could just mass mail a bunch of vinyl shops that list of questions and fish them for more information.... we learn a lot from them and get a better picture how they compare to each other!
Last edited:
Questions are in black, answers in BLUE Marked up my comments in green below (Luckythedog):
1) Levels.
We want a 1kHz reference tone cut at 5 cm/s. At what level should we set this reference in our digital files?
For a full frequency sweep, what is a recommended level? We don’t want to over drive the cutter at high frequencies .
There is no fixed connection between digital level in audio files and analog (mechanical) level on record. We always set cutting level according musical content.
If we know the reference level is supposed to be 5 cm/s we set to cutting level to meet this requirement.
5 cm/s is quite low for reference tone. It corresponds to -9 dB (= 16 µm peak to peak amplitude) of Neumanns reference level at 1 kHz (0 dB = 14,14 cm/s = 45 µm PP).
For real time cutting current, temperature and acceleration are limiting so -20 dB of Neumanns level is safe. For non real time cutting (half speed, 2/3 speed, 3/4 speed) mechanical parameters (groove angle, groove radius, ...) are limiting. In other words it is possible to cut groove that is impossible to play at least without heavy distortion.
To prevent this problem I recommend to use 45 rpm and cut frequency sweep at outer side of the disk from higher to lower frequencies.
I take this to mean that so long as we are specific about what level in the digital master file = 5cm/s, their mastering house can work with that. The comment that 0dB= 5cm/s is quiet is good, and their figures suggest plenty of headroom for cutting. We need to end up with the test velocities we require on the record, and we don't care or need to worry about lathe levels so long as our test tracks don't exceed any limit. Which looks good, given the figures - we get up to about 24cm/s peak as things stand. Lathes are geared to work with programme material having normal crest levels, and presumably set to operate at 20dB headroom by the sound of it.
As to frequency sweep test levels, one mistake over generations past has been to think playback f response is not level sensitive. So the test record as it stands has multiple tracks at different levels. If one arranges for test tracks to stay within playable limits, and be RIAA compensated, the frequency range has to be restricted at higher levels. That's why the candidate track list has several levels and sweep ranges, but they all should be cuttable and playable IMO, based on what is said here.
Vinyl playback isn't really 'frequency' limited, it's limited by angles and curvature which depend on level. However, mastering probably whacks programme material through a brick wall filter that isn't level sensitive - so we'd need to turn this off to get up to 50kHz, and mastering engineers would then need to know the test tracks aren't going to turn the cutter head into crispy fried duck.
2) EQ and compression.
Of course we don’t want EQ or compression for any files. These are test tones straight cut.
Can we apply our own RIAA curve to the files and you turn off yours? Or does this cause you problems?
For test LP we don't use any processing except cutting level settings. There must be clear information about test records and requests in tracklist. To help orientation different width of transition grooves can be used to separate specific signals.
We can switch off upper time constant (2122 Hz) only. There is no benefit if you apply your own RIAA curve. There are several other circuit in the signal path and RIAA curve has to be last one!
I'd assumed this was the case, and we'd need the master file not to be pre-compensated, but the grooves cut would be RIAA compensated. So when it comes to specific required groove shapes, other than sines, we need to work with that in the master file and it will be fine.
3) Frequency Response
What are your frequency limits with pure tones and sweeps? We’d like to get all you are capable of cutting and hope to go above 20 kHz.
Is it advisable to use 1/2 speed mastering?
Generally we are able to cut more than common pickup is able to play. Usually geometric parameters are most limiting, so the best results are achievable for 45 rpm and outer side where should lie most critical signals. Differences between cutting and playback stylus cause several nonlinear distortion, crosstalk up to -6 dB, left and right channel time shift and many more. It depends what you prefer. I did frequency sweep cut till 30 kHz at -16 dB. Frequency response for the SAS playback stylus was almost flat till 25 kHz.
1/2 speed mastering lower current and temperature 4 times so it is no longer problem. On the other hand it causes problems for cutting low frequencies. Sometimes I rather use 2/3 or 3/4 speed.
Frequency response depends upon level, because it's limited by cut angles and curvatures which depend on level. Cutting is also limited by mechanical power in the head, which is a function of level as well as frequency.
I hope, so long as programme levels are small eg -54dB ref 5cm/s, we might be able to stretch to 50kHz if mastering engineers are willing to try. It would be very useful to have those tracks. If they can cut 30kHz at -16dB I think 50kHz at -54dB or -36dB might be possible.
4) File format.
Is there any real advantages to 192K bitrate, or is 96K enough?
I believe you won't hear the difference even for critical test signals. I measure ears to our every new cutting engineer. The absolute record is 21 kHz, but most people are not able to hear more than 18 kHz. It depends on age and many others factors.
It is TRUE that design of antialiasing filter is easier for higher sampling rates. On the other hand design of the DA converter clock is harder for higher frequencies. It depends on DA converter. We use Prism Sound Orpheus DA coverter for cutting.I agree - I doubt bitrate matters.
5) Hole centering. How precise can it be? It’s important for a test disc.
Hole centering together with hole size tolerance and turntable center spindle tolerance also are responsible for eccentricity error. For musical signal eccentricity below 200 µm is acceptable and it not cause problem. For test harmonic signal you can always measure some frequncy fluctuation.
Center hole tolerance is 7.24-7.33 mm.
Hmm. That is a BIG problem. Tolerance needs to be at least twice as tight if the tracks are to be used with polar plot software, and be able to know what is record and what is other pitch variation. This was the original reason to do the test record, and even if the records were made to this spec, it would be a fail, and no better.
So something special needs to be done, methinks, since we consider it critical............
6) Is there anything you, as a mastering engineer, would like to see on a test LP? Something you think should be included or can be done better than on other test LPs?
I would consider signal for measuring polarity, crosstalk, pink noise, arm resonance, rotation fluctuation, intermodulation distortion, pickup azimuth settings, and many more.
There are many things (cutting lathe, head and stylus selection, non real time cutting, cutting head protection bypass, highly accurate cutting & feedback level settings, endless & closed groove programming, distortion precompensation),that could be done better but they requires additional time to prepare cutting and we charge extra cost for them.
We can create test record according to detailed description only. We use high accuracy 4 tone signal generator special invented for vinyl records.
Here are basic features:
We want a 1kHz reference tone cut at 5 cm/s. At what level should we set this reference in our digital files?
For a full frequency sweep, what is a recommended level? We don’t want to over drive the cutter at high frequencies .
There is no fixed connection between digital level in audio files and analog (mechanical) level on record. We always set cutting level according musical content.
If we know the reference level is supposed to be 5 cm/s we set to cutting level to meet this requirement.
5 cm/s is quite low for reference tone. It corresponds to -9 dB (= 16 µm peak to peak amplitude) of Neumanns reference level at 1 kHz (0 dB = 14,14 cm/s = 45 µm PP).
For real time cutting current, temperature and acceleration are limiting so -20 dB of Neumanns level is safe. For non real time cutting (half speed, 2/3 speed, 3/4 speed) mechanical parameters (groove angle, groove radius, ...) are limiting. In other words it is possible to cut groove that is impossible to play at least without heavy distortion.
To prevent this problem I recommend to use 45 rpm and cut frequency sweep at outer side of the disk from higher to lower frequencies.
I take this to mean that so long as we are specific about what level in the digital master file = 5cm/s, their mastering house can work with that. The comment that 0dB= 5cm/s is quiet is good, and their figures suggest plenty of headroom for cutting. We need to end up with the test velocities we require on the record, and we don't care or need to worry about lathe levels so long as our test tracks don't exceed any limit. Which looks good, given the figures - we get up to about 24cm/s peak as things stand. Lathes are geared to work with programme material having normal crest levels, and presumably set to operate at 20dB headroom by the sound of it.
As to frequency sweep test levels, one mistake over generations past has been to think playback f response is not level sensitive. So the test record as it stands has multiple tracks at different levels. If one arranges for test tracks to stay within playable limits, and be RIAA compensated, the frequency range has to be restricted at higher levels. That's why the candidate track list has several levels and sweep ranges, but they all should be cuttable and playable IMO, based on what is said here.
Vinyl playback isn't really 'frequency' limited, it's limited by angles and curvature which depend on level. However, mastering probably whacks programme material through a brick wall filter that isn't level sensitive - so we'd need to turn this off to get up to 50kHz, and mastering engineers would then need to know the test tracks aren't going to turn the cutter head into crispy fried duck.
2) EQ and compression.
Of course we don’t want EQ or compression for any files. These are test tones straight cut.
Can we apply our own RIAA curve to the files and you turn off yours? Or does this cause you problems?
For test LP we don't use any processing except cutting level settings. There must be clear information about test records and requests in tracklist. To help orientation different width of transition grooves can be used to separate specific signals.
We can switch off upper time constant (2122 Hz) only. There is no benefit if you apply your own RIAA curve. There are several other circuit in the signal path and RIAA curve has to be last one!
I'd assumed this was the case, and we'd need the master file not to be pre-compensated, but the grooves cut would be RIAA compensated. So when it comes to specific required groove shapes, other than sines, we need to work with that in the master file and it will be fine.
3) Frequency Response
What are your frequency limits with pure tones and sweeps? We’d like to get all you are capable of cutting and hope to go above 20 kHz.
Is it advisable to use 1/2 speed mastering?
Generally we are able to cut more than common pickup is able to play. Usually geometric parameters are most limiting, so the best results are achievable for 45 rpm and outer side where should lie most critical signals. Differences between cutting and playback stylus cause several nonlinear distortion, crosstalk up to -6 dB, left and right channel time shift and many more. It depends what you prefer. I did frequency sweep cut till 30 kHz at -16 dB. Frequency response for the SAS playback stylus was almost flat till 25 kHz.
1/2 speed mastering lower current and temperature 4 times so it is no longer problem. On the other hand it causes problems for cutting low frequencies. Sometimes I rather use 2/3 or 3/4 speed.
Frequency response depends upon level, because it's limited by cut angles and curvatures which depend on level. Cutting is also limited by mechanical power in the head, which is a function of level as well as frequency.
I hope, so long as programme levels are small eg -54dB ref 5cm/s, we might be able to stretch to 50kHz if mastering engineers are willing to try. It would be very useful to have those tracks. If they can cut 30kHz at -16dB I think 50kHz at -54dB or -36dB might be possible.
4) File format.
Is there any real advantages to 192K bitrate, or is 96K enough?
I believe you won't hear the difference even for critical test signals. I measure ears to our every new cutting engineer. The absolute record is 21 kHz, but most people are not able to hear more than 18 kHz. It depends on age and many others factors.
It is TRUE that design of antialiasing filter is easier for higher sampling rates. On the other hand design of the DA converter clock is harder for higher frequencies. It depends on DA converter. We use Prism Sound Orpheus DA coverter for cutting.I agree - I doubt bitrate matters.
5) Hole centering. How precise can it be? It’s important for a test disc.
Hole centering together with hole size tolerance and turntable center spindle tolerance also are responsible for eccentricity error. For musical signal eccentricity below 200 µm is acceptable and it not cause problem. For test harmonic signal you can always measure some frequncy fluctuation.
Center hole tolerance is 7.24-7.33 mm.
Hmm. That is a BIG problem. Tolerance needs to be at least twice as tight if the tracks are to be used with polar plot software, and be able to know what is record and what is other pitch variation. This was the original reason to do the test record, and even if the records were made to this spec, it would be a fail, and no better.
So something special needs to be done, methinks, since we consider it critical............
6) Is there anything you, as a mastering engineer, would like to see on a test LP? Something you think should be included or can be done better than on other test LPs?
I would consider signal for measuring polarity, crosstalk, pink noise, arm resonance, rotation fluctuation, intermodulation distortion, pickup azimuth settings, and many more.
There are many things (cutting lathe, head and stylus selection, non real time cutting, cutting head protection bypass, highly accurate cutting & feedback level settings, endless & closed groove programming, distortion precompensation),that could be done better but they requires additional time to prepare cutting and we charge extra cost for them.
We can create test record according to detailed description only. We use high accuracy 4 tone signal generator special invented for vinyl records.
Here are basic features:
- 64 bit floating point
- cutting lathe rotation synchronization
- 1 ns signal period settings resolution
- 0,000001 dB level settings
- 10 nm amplitude settings
- 4 simultaneous signals
- special signals (pink and white noise, polarity, impulses, linear and logarithmic sweep, trapezoidal & triangle groove shape)
- modulated signal
- constant amplitude sweep
Last edited:
I've seen mistracking damage album grooves at the inner tracks when that same cartridge/arm combo doesn't exhibit mistracking damage at the outer grooves. You can pretty clearly hear the damage, and it's fairly common if you start chattering on the last track.
I've seen situations where the entire collection of LPs shows inner groove damage, save for a few albums that are not demanding and not favoured so low play count, because the owner insisted on pushing the lower limit of the tracking force spec and was, to be honest, deaf, despite the existence of the HiFi in the room.
Lucky, this still bothers me. Several of the CBS Labs disks don't work through an RIAA preamp and you need to build a flat/special preamp. My gut feeling is that normally a mastering engineer would not even consider anyone would want that.
He mentions "triangular groove shape" on the CBS square wave test disk they are true triangular grooves. I'm still not understanding if his system applies the predistortion or if we need to or in fact if we actually are going to submit tracks or need to specify the inputs to his special signal generator. Another one is the constant velocity frequency sweep (no RIAA means it needs to start at 500Hz), without RIAA pre-emphasis the extreme highs might be easier on the system. I mean torturing the lathe because the user is going to roll the output off sort of makes no sense.
Last edited:
I have not seen this, but possibly the mistracking from a bad VTA at the correct tracking force is not immediately as damaging. We were not fine adjusting VTA by varying the tracking force but actually stopping and raising and lowering the arm.
I always set tracking force at the high end of the recommended range, no lower than the mid point at least. That is a starting point, that could be changed based on how everything works, so there have been cases with certain cartridges + tonearms where it ends up on the mid to low side, but that's not the starting point.
Too low a tracking force is far more damaging than too high (if there is such a thing as too high, within the manufacturer's recommended range, which I doubt).
In truth I see no evidence that tracking forces up to 3 gm cause any excessive wear at all, on the vinyl or the diamond. But there is a tendency I've observed hundreds of times where people set tracking force on the low side of the recommended range, which they feel means reduced record wear. They are wrong.
Somewhat more on topic, thanks for the Q+A post(s). Very informative, and in language I can understand a bit better than some of the super-technical posts (which are welcome, they are just beyond my pay grade).
Last edited:
Many good points above. Since the RIAA mastering curve (AKA, reverse RIAA) is going to push the top end up by 20dB, it's good to be careful up there is cutting loud HF signals. I wouldn't think this is a problem for cutting normal music files, as they tend to roll off pretty fast up top. As Lucky notes, his sweep files take this into account with level reduction.
I could ask how hot they can safely cut a 40 or 50 kHz signal and we set the sweeps at that and lower. Using 2/3 speed mastering should help.
Scott I understand your point of not pushing the top end, simply to attenuate it again in the listener's RIAA preamp. But there are advantages to having a sweep that plots flat after RIAA EQ. It's easier to read, and can be done with any normal phono preamp. GZ does note that they can turn off the top end of the curve - but is it useful to do so?
I could ask how hot they can safely cut a 40 or 50 kHz signal and we set the sweeps at that and lower. Using 2/3 speed mastering should help.
Scott I understand your point of not pushing the top end, simply to attenuate it again in the listener's RIAA preamp. But there are advantages to having a sweep that plots flat after RIAA EQ. It's easier to read, and can be done with any normal phono preamp. GZ does note that they can turn off the top end of the curve - but is it useful to do so?
I think it is important to differentiate distortion from incorrect groove/stylus geometry and the stylus leaving the groove and smashing back in. They might not sound that different in casual listening, it would be interesting if one could learn the difference.
I think it is worth mentioning the other approach and asking more questions. Remember constant velocity translates to flat for a velocity sensitive cart and flat preamp. CBS dealt with this by having constant amplitude sweeps below the 500.5Hz RIAA time constant and constant velocity above. This essentially eliminates most of the worries about the two RIAA curves being perfectly complementary.
It's OK - the candidate tracklist as put forward is intended to be cuttable and playable without customisation except for turning off compression/limiting, turning off any 20kHz brickwall filter, turning off acceleration limiting, and cutting some lock grooves. Which I understand to be the sort of things GZ said they would be willing to do, for a fee.
IMO it's better for the f sweeps and noise tracks to be normal RIAA compensated so that they play back flat on normal preamps. As to whether sweeps should be constant velocity then, I think not. But all of these things are within our grasp, if we choose, working with mastering RIAA on - just by precompensating our digital masters. Likewise we can still make any shape groove, so long as we know which way the mastering will be done: RIAA on or off. Also need to know how the mastering RIAA behaves above 20kHz: whether there is a shelf/4th time constant.
I think the big problem is hole size/centring. Maybe there could be 'standard' version with standard production tolerance, for those not interested in pitch stability polar plot SW. And a 'plus' version, which is made without the hole then sent to a machine shop to have the hole accurately made ? And/or a delux version without a hole which one can make oneself ?
LD
Last edited:
This also allows sizing the hole to the spindle on your own turntable.
These can vary by more than 0.1mm.
