Theoretical Record Frequency Response

[Craig Maier]

Quoting Doug,

"Craig,
Now that you've pointed out the absolute best theoretical frequency response on a 78rpm recording...is there a way to flaunt this fact in the INF and other filter? obviously anything above ~8kHz is artefact."

------------------------------

Doug,

I think that you just did. Lol!

Anyway, I think that there may be some distortion products that may exist above the theoretical calculations that I performed. Some folks may be interested in keeping those products. And, keep in mind, that roughly 8 KHz (7.7 KHz) response only occurred at the very end of the record. So, if you do not mind the variability factor, some may want to operate at a higher bandwidth on some recordings wherein the master did not account for this situation.

[Dan McDonald]

You'd think that the producers might sometimes make use of better frequency response in the beginning - with a violin solo, cymbals or something that produced high frequency response. I don't know if they did, but I'd imagine that the good ones did take care to make the record sound the best it could.

[Craig Maier]

It looks like the Edison Company may have been aware of this phenemenon. Check out this thread on the forum:

http://www.diamondcut.com/vforum/showthread.php?t=2104

Since the human ear will respond to the power spectral density of material, I wonder if the "weak centers" complaint was really just a fall-off in the frequency response of the recordings?

[Dan McDonald]

And wasn't that part of the argument that Edison made for the superiority of the cylinder in the early days? I don't have a reference, but it seems to me I read that somewhere.

Dan

[Craig Maier]

Clearly the constant tangental velocity of a cylinder offers that benefit. Although I do not have the reference at hand, I do recall in my readings that Edison embraced the cylinder over the disc format early on because of this situation, among others like constant stylus tracking angle, etc.

As you may know, Edison patented his phonograph including a media that included:

cylinder
disc
moving tape

He also patented -

vertically cut
laterally cut

Yet, he chose to commercialize vertically cut cylinders; I believe that was done for performance reasons.

[Dan McDonald]

For 78s, I'd be tempted to make sure they were all limited to the 8kHz at the beginning. If not, I would be tempted to use the starting theoretical frequency and then try to rework the record so that the ending was closer to the start, than to just get rid of what was available at the befinning of the record.

That is, I think I'd look at the spectrogram or the spectrum analyzer and try to come up with a multifilter that could make the back end look and sound like the front end, rather than the other way around. Of course, if the engineers already 'lopped off' the front end, then there's nothing to change.

Dan

[Craig Maier]

Perhaps a good tool to deal with this situation (when the beginning response is not "lopped off" by the record manufacturer) might be the "Filter Sweeper" set for Low Pass. Since this is a "gentle" slope filter system, I would recommend trying the following:

* Start Freq: 6 KHz

* End Freq: 12 KHz

* Linear Response

ps - I have not tried it yet.

[Doug]

I just mentioned it since most of the records I've seen from 1920s-40s have the audio lopped off ~8kHz. Have seen some as low as 4kHz! (constant across the entire record).

I have a few from the 50s where they do have a frequency range up to ~ 16kHz (including a few MGM recordings) - but most are not recorded like that...for Aussie purposes anyway.

[Craig Maier]

Decca ffrr

I believe that Decca advertised that their ffrr process produced up to 15 KHz or 16 KHz of top end bandwidth on their 78's. What we do not know is where on the record that response was realized. If they cut the master shallower than normal, then the player stylus would ride higher on its skirt. A narrower profile of the player stylus would be in contact with the groove wall. Thus, effectively, the stylus would appear to be smaller than advertised from a physics perspective. That would permit higher frequency response values to be achievable compared to the theoretical ones shown at the top of this thread. Other record companies may have followed suit.

So, it is certainly a possibility.

[Doug]

well...this may be my last question on this topic:
Assuming that the 78 has a frequency range up to 8kHz...is it better to have a steep low pass filter at 8kHz or a slow low pass filter at 6kHz?

I notice you often suggest the 6kHz 6db/8Ve low pass filter Craig - is that personal preference?

[DougMac]

Excellent Question

Assuming that the 78 has a frequency range up to 8kHz...is it better to have a steep low pass filter at 8kHz or a slow low pass filter at 6kHz?

I notice you often suggest the 6kHz 6db/8Ve low pass filter Craig - is that personal preference?

Personally, I use a low pass filter set around 7kHz that's steeper, either 12 or 18 db per octave. On some source material, I use 24 db/octave.

This is done by ear, I guess that's where the "artistry" comes into restoration. I'm curious, though, to hear Craig or others address the pros/cons of using fairly steep rolloffs.

Doug
PS To avoid confusion, maybe the other Doug should sign "Aussie Doug" and I should sign "Georgia Cracker Doug"

[Craig Maier]

My only concern with steep slope filters is that they exhibit a high "Q" at the break (or corner) frequency. This can create sort of a "whistleing" sound (due to "peaking") or resonance at that frequency on some musical material. As a matter of fact, it is worse with the Chebyshev response filter compared to a Butterworth. If you want to hear this effect, create a "Make Waves" noise file. Experiment with LPF slopes up to 24 dB / Octave at 7 or 8 KHz on that file. Then switch between Chebyshev and Butterworth and you should be able to hear what I am trying to put into words. If you want to see this effect, bring up the spectrum analyzer and set it for:

4096 FFT's
5.38 Hz Resolution
Averaging Mode
Blackman Window
50 dB Range

Preview the file with the system set for 24 dB/Octave, Chebyshev and watch the ripple build up at the corner frequency of the LPF.

But, if this high "Q" whistle is not noticable or annoying, then the benefit of a steep filter can be obtained. The above-band noise rejection will be greater if filters with steeper slopes (and/or Chebyshev response) are used.

[Craig Maier]

Theoretical Frequency Response of Blue Amberol Cylinder Records

I became curious as to what the Frequency Response of Blue Amberol Cylinder recordings could be which were developed around 1912. Here is my analysis based on the equation(s) established at the beginnning of this thread:

F = (Pi x D x RPM) / (W x 120)

wherein

D = Usable Diameter of the Record (which is constant with a cylinder) (in inches)

RPM = Revolutions Per Minute of the Record in Question

W = Smallest dimension of the styli (in inches)

Pi ~ 3.1416

-----------------------------------------

Now, lets calculate the maximum theoretical frequency response of an Edison Blue Amberol Cylinder

This type of record is 2.125 inches in diameter (by inspection).

RPM = 160 (from the Diamond Cut users manual)

The Diamond Cut Users manual calls for the use of a 3.7 to 4.2 mil spherical stylus for this type of recording. So, using a spherical 3.7 mil phono stylus (the best case), the frequency response will be as follows:

Theoretical Starting Frequency Response: 2.407 KHz
Theoretical Ending Frequency Response: 2.407 KHz
Response Variance: 0 KHz
Average Response: 2.407 KHz

Conclusion: This format was inferior in terms of frequency response compared to its pre-existing rival, the 78 RPM Disc record. See the top of this thread for details. However, to be noted is that the frequency response is a constant during play, unlike that of a disc record (lowest value around 7.7 KHz). I guess one could say that it was consistently poorer than its disc counterpart - - - probably another reason for this formats ultimate demise.

Note: I presented the data in the same format as was done for the disc records for consistency.

[Dan McDonald]

Hey Craig -

What role does the medium play? I would think the comparison for the cylinder innovation would be to earlier formulations. Probably a lot better than wax, right? Although some of the wax ones sound fairly decent.

Dan

[Craig Maier]

I am not certain, but I would think that the medium would have more of an effect on the noise floor and the records durability than its frequency response. On the other hand, the materials that the master had been made out of could effect frequency response due to frictional effects on the cutter lathe stylus system. Softer materials used for the master would offer less friction and thus a flatter frequency response up to the theoretical limit based on cutter stylus width.