Optimum settings for recording LPs with further editing (sample rate and bit depth)

[Craig Maier]

Here is a thought along the lines of higher sample rates. Instead of using 96 kHz, consider using 88.2 kHz if your final goal is to create CDs which are 44.1 kHz. The advantage of this would be the rational division of 88.2 kHz down to 44.1 kHz which is a simple divide by two having no remainder, thus no truncations or roundings. And, I am guessing that the small difference in bandwidth capability between 96 kHz and 88.2 kHz will not have a significant impact on your de-noising results. Your Diamond Cut software supports 88.2 kHz recording (the "custom" field) and file conversions from 88.2 kHz to other values too.

Craig

[cromag]

I picked up a 2 TB external hard drive last summer for about $60.

Last fall there was catastrophic flooding in Thailand which, according to news reports, knocked out 40% of the worldwide hard drive manufacturing capacity. Prices went up initially due to reduced supply, then went up even more when computer manufacturers bought up as much inventory as they could.

Prices seem to be coming back down ... slowly ... but they're still much higher than last year.

[Craig Maier]

Hi Cromag,

That is interesting to know - - - wasn't aware of the impact due to the situation in Thailand. Do you know what 60 dollar hard drive that you purchased last summer for costs today?

Craig

[cromag]

My memory was a bit faulty. I checked my invoice and I actually paid $75 for it. I'm still happy.

The exact model I have seems not to be available anymore at my usual suppliers. I don't remember it being a clearance special, but if that was the case it might have reduced the price slightly. I did find a couple of vendors that are listing it as still available, and the price range seems to be $120 to $180.

The same manufacturer's current equivalent model -- 2 TB, USB 2.0 external drive -- is listed at several computer supply sites at about $150 to $200. I checked the manufacturer's website and this model is now listed as recently discontinued, so I don't know if prices will be affected in the coming weeks.

The same manufacturer is offering a bare 2 TB drive (with faster specs, presumably for use as an internal drive) for $200, so you wouldn't save anything by using your own housing.

[Marc Hildebrant]

With regard to sampling rates......

I do not disagree that high frequencies can help the filters to find impulses, but, I'm not sure that the current audio preamps with discrete RIAA components pass much energy above 20 KHz.

Could it be that the new "flat" preamps are needed to pass the ultrasonic signals through for 96KHz benifit ?

Marc

[Craig Maier]

Hi,

I guess it has a lot to do with the particular RIAA preamp. Nothing is really defined above 20 kHz as far as I can recall in the RIAA standard. So, an RIAA preamp could keep rolling off at a constant slope above that frequency or it could also "shelf" above 20 kHz and be in compliance. One can certainly design an RIAA preamp that complies and still has a "shelved" response above 20 kHz. I am not sure what the general practice is. An RIAA preamp that I designed and built around 30 years ago or so (it has 9 x 12AX7s - 18 triodes in total), shelves above 20 kHz and keeps on going up to around 200 kHz or so, so that one would be ok in the circumstance that you describe. To be sure and as you state, the "flat preamps" are flat to within 1 dB from 10 Hz up to 150 kHz (CTP - 1000) as I recall. I guess if I were you and had an unknown RIAA preamp, I would take it and feed it with random noise using Diamond Cut through a 40 dB attenuator from the line out of my soundcard. I would use a second instance of Diamond Cut to bring the output signal from the preamp back into the soundcard line input and use the DC Spectrum analyzer to plot its frequency response above 20 kHz. Just be sure that the file that you generate using the Make Waves Generator is using at least 96 kHz sample rate so that there are signals up to 48 kHz available to stimulate the DUT.

ps - T pad attenuator tables of values can be found in the Diamond Cut users guide which may be helpful. Normalize the PAD for 50 kOhms Input and Output Z values.

[Craig Maier]

Hi Again,

As I think about your question a bit more, I must say that unless the RIAA preamp "brick - walls" above 20 kHz, there will be some useful signal energy above 20 kHz. It would be best for the purposes of the discussion if the preamp "shelved" above that frequency, but even if it falls at a constant slope, it is only a little more than an Octave from 20 kHz up to 48 kHz where the sound card would brick wall if one uses 96 kHz sampling rates. And, it is hard to imagine that each Voltage gain stage in any given audio preamplifer would have the proper pole alignment to create a "brick wall" low pass filter response of high order.

In other words, I think that most RIAA preamps will still pass useful signal energy past 20 kHz. I hope that makes sense. To prove it on your specific preamp, consider using the test that I outlined above.

Craig

[Craig Maier]

Hi Again,

Sorry to keep popping up - - - but I remembered some useful supplimentary information pertaining to the RIAA curve which you can use to test any RIAA phono preamplifier for accuracy. You can also use this with a flat preamp in conjunction with the Diamond Cut Virtual Phono Preamp to see its accuracy (basically it measures perfectly), or to measure a time constant based preamp (which will fall short, but the inaccuracy will depend on the quality of the unit).

Goto the helpfile in your DC8 or Forensics8 software and search the index for:

RIAA Curve Table of Values

It lists all frequencies from 20 Hz to 20 kHz and the relative gain of the system across the audio band, referenced to 0 dB @ 1 kHz. Those of you with flat preamps and the Virtual Preamp will be quite amazed at the accuracy of the system. Those with regular time constant based preamps can easily see the degree of accuracy provided by their preamps by excititing it with the Make Waves White Noise Generator. Use the Spectrum Analyzer to measure the response to the White Noise excitation.

Craig

[Dan McDonald]

Craig -
It seems like I'd need to set up some guidelines for the experiment. I might have time to run some of it this weekend.
It looks like the best way to do this would be to use the EZ impulse at certain presets and see what each preset finds?
So one of the lp presets and one 45 preset and one 78 preset? it's turning into a big design.

[Craig Maier]

Hi Dan,

I think that the test is more subjective than that. Just finding impulses and using the quantity as an indicator will not truely reflect the difference in performance between methods or techniques. So, it not only depends on the number found, but how accurately they are replaced. So, I think that the test would involve two differeing procedures and then a subjective comparison of the results. I would adjust the controls optimally for each test (because the optimal settings will be different). But, the final results may vary revealing a difference in the performance of one technique over the other.

For example, one could transfer the same record track at 44.1 kHz and also at 96 kHz (just as an example). Then, do the best cleaning up each one independently. When done, go back and compare the final outcomes subjectively. I know it does not sound real scientific, but I think that ultimately it is the sound quality that counts, not the numbers.

Make sense?

Craig

[Dan McDonald]

Yes - it makes perfect sense, but it seems like we'd have to have a bunch of people doing it so that we would know whether there was bias (I might try harder with 96kHz or something like that). Well, anyway, I might do some test transfers soon and see what I can do with them.

[Craig Maier]

hmm - yes, it is a very biased test, indeed. Maybe to improve the test, one could do the best that they can with set of test conditions and have a third party (girlfriend, wife, kids, etc) decide which sounds the best. That still not dial out the bias of the restorer.

As I recall, one of the proposed tests is to keep the sample rate at 96 kHz and transfer at 16 rpm vs 33.3 rpm and see if that makes a difference. In theory, the 16 rpm transfer to 96 kHz should work better than the 33.3 rpm transfer, but a test is the true arbitor.

Craig

[Craig Maier]

This thread has really arroused my interest in the sample rate questions. I just ran a sample rate test on my best soundcard. I will not say who made it, but it is a well known brand and model. I tested the bandwidth that it has at 44.1 kHz, 96 kHz and 192 kHz sample rates. I used the Diamond Cut Forensics8 v 8.1 to do the tests (spectrum analyzer).

The result was very unexpected. At 44.1, the brick wall occurred at around 21 kHz (expected). At 96 kHz, the brick wall occurred at around 47 kHz (expected). But here is the clinker - - - at 192 kHz, where do you think that the brick wall is? Well, (little drum roll needed) it was at 21 kHz!!

Wow. I can not imagine why they did that. But, they did. So, it would appear that one needs to be careful when just assuming that higher is really higher. 96 KHz would be a much better choice than 192 kHz and 192 kHz produced the same performance as 44.1 kHz, but eats disc space at 4 times the rate with zero relative benefit with this particular sound card.



Craig

[sus4chord]

Craig -

Does your soundcard actually support 192? I know that one of my "well known" cards only supports up to 96. Just curious.

[Craig Maier]

Hi John,

I am not really sure about that. It is around 4 years old and I used Diamond Cut Forensics8 v.8.1 (latest version) to make the test recordings. The system accepted the 192 kHz recording sample rate value and it created a file having that value as indicated by the Diamond Cut File Information feature (View Menu). But, I guess that does not necessarily mean that the sound card supports 192 kHz - - - maybe it just emulated it. But, if it just emulates 192 kHz, I wonder why it does not down-sample to 96 kHz since it absolutely does handle that situation correctly. My feeling on the matter is that if it does not support a certain sample rate, it should not have allowed the recording at that rate, but who knows? Maybe my expectations are unusual?

Craig