Yes, I can imagine conventional instrumentation having problems here. Did you catch my direct 3,300,000 point plot of the STR130 1min 40-20k sweep? I'm sure there are other techniques but I got a fairly good result with the brute force one. Unfortunately the analog generators that they must have used had some glitches.
Agree on some locked grooves for continuous tests. Chardas has some, anyone have this?
Could possibly divide into absolutely necessary and bonus disk. For instance frequency response at a standard level vs. the extra level tracks. Also remember if we could get a total commit to 1000 pieces (sans labels/artwork) the price becomes far below any of the best disks out there.
Doing online research for what to use for testing I came across this thread in a Harbeth forum: Speech test recordings - where to obtain? - Harbeth: For 40 years - the world's most natural sounding loudspeakers
Would it make sense to include a Speech intelligibility test recording? On the bonus disk?
Would it make sense to include a Speech intelligibility test recording? On the bonus disk?
Hi Pano,
-Chris
I don't have a problem with that. I just didn't want it to turn into a 3 ~ 4 disc set. Also, you have to watch for turntables that use acceleration to sense arm pick-up. So the speed that the arm swings must be within normal limits. I'm sure that the extra room can be taken up by extending the duration of some tones, especially the W&F / speed track. I would like to see that and the THD tests grow to take up any extra time we might end up with on that side. Some tests can take a little while, long if you have old test equipment or manual THD analysers. Many instruments have an internal auto-cal sequence every so often.
-Chris
Just tossing around a random idea... Would a "multisided" side with several grooves per side make sense in any application? Like different tracks at different frequencies from outside all the way to the spindle, all on the same side?
Edit: You could probably even do 33 and 45 tracks on the same side that way.
Edit: You could probably even do 33 and 45 tracks on the same side that way.
Last edited:
The issue isn't one of instrumentation: it's that vinyl playback involves several physical mechanisms and transducers which are prone to be level sensitive as to frequency response.
So no single level sweep, such as the STR130 sweep, can be particularly meaningful or definitive no matter how it's analysed.......the playback f response might vary significantly with level of the sweep test, over the dynamic range likely to be seen in programme material. Same is true for spectral noise tests, of course.
It's a significant non-linearity peculiar to vinyl that isn't characterised or discussed AFAIK, mostly because there aren't test records to examine it............
LD
I might be overthinking it but I was considering the problem of simple level detection in the presence of large amounts of distortion. For instance I've seen sine waves turned nearly into tri-waves (especially at high frequencies), so what is the correct amplitude of the response there?
Exactly. What you've very probably seen is the result of slew rate limiting: in the first instance a sine wave becomes more triangular, and progressively a triangular wave loses amplitude as slew rate becomes more limited.
Peak slew rate of a sine wave depends on its amplitude, as well as its frequency, of course.
So, if one measures f response by looking at fundamental amplitude (as is the norm) the outcome would depend on test level in the event of significant slew rate limiting.
Unfortunately, several significant mechanisms contribute to slew rate limiting in vinyl playback: mostly via various mechanical bending and magnetic loss mechanisms. It also crops up in speakers and mics, FWIW.
AFAIK, the industry has always ignored the fact that f response might depend upon absolute level, and seems the semi-tech public has pretty much been brought up on an unstated (and often false) premise that f response measured at one level is adequate and applicable at all levels.
Anyways, this is (AFAIK) fairly uncharted water, and if we (the people) are doing a test record it seems easy enough to includes f sweeps and spectral noise at various levels across the dynamic range of playback, to enable proper measurement of some of the playback mechanisms involved, methinks.
LD
That would be a strength (or weakness?) of the direct Hilbert transform method which yields only the level of the fundamental. I do think it does it with more rejection of harmonic content than some pure analog techniques.
Slightly daft but hopefully relevant question. I've noticed some reviews publish lateral and vertical distortion measurements for cartridges, but vertical only at 315Hz or so as that is all the test record has. Now given that we have +/-45 generators we don't necessarily need vertical, but possibly lateral, left channel and right channel sweeps at different levels.
Makes sense or am I being confusled again?
Makes sense or am I being confusled again?
This may be a pie in the sky idea but I have at my disposal a cnc engraving machine with
Hydrostatic linear guideways (X & Y)
Linear motors ( not servos! )
Sony 0.4um accuracy scales (per 420mm) - resolution 138nm
Precision 40Krpm spindle
I could in theory buy a diamond tipped cutter and machine master test LPs. Hunting around for suitable PCM/AIFF to groove software....
The linear motor controllers are by YASKAWA/OMRON and have vibration suppression feedback from the scales to make sure the cutter is where it should be. Clever stuff. Video here.
Vibration Supression
Thoughts?
Hydrostatic linear guideways (X & Y)
Linear motors ( not servos! )
Sony 0.4um accuracy scales (per 420mm) - resolution 138nm
Precision 40Krpm spindle
I could in theory buy a diamond tipped cutter and machine master test LPs. Hunting around for suitable PCM/AIFF to groove software....
The linear motor controllers are by YASKAWA/OMRON and have vibration suppression feedback from the scales to make sure the cutter is where it should be. Clever stuff. Video here.
Vibration Supression
Thoughts?
Last edited: