There is a lot to do about MQA re-releases
that have limited bandwidth despite the high sampling rate.
This might have to do with the properties of analogue recorders, as I mentioned in my
video "Why some MQA 192 kHz files stop at 20 kHz".
So let's look at some analogue tape recorder properties for better understanding.
Almost all analogue recordings made since the second world war are made on tape recorders.
They work by magnetising a magnetic pattern on a 'plastic' tape covered with ferro oxide -
in essence a sophisticated form of iron rust.
Without going into detail too much, I would like to explain those artefacts that are typical
of analogue tape recorders.
Let's start with the recording.
The tape is first pulled along a very simple erase head, in essence an electromagnet that
is fed with a high frequency signal.
This shakes up the magnetic fields that might already be on the tape in a random way,
so erasing the tape.
Then it passes the record head.
This also is an electromagnet but far more delicate.
This head creates a magnetic field analogous to the electric signal that presents the audio signal.
This varying magnetic field creates a variant magnetic field in the iron oxide on the tape.
Unfortunately the magnetic field radiated by the head is wider at low frequencies.
Therefore the track on the tape will be wider for low frequencies.
Later, tape heads got poles in the shape of a butterfly to reduce this effect but these
butterfly heads were not fitted on all tape recorders for they were only suited for stereo
recording while for instance time code - for use with video - was not possible.
The Butterfly heads had another advantage: they could write wider tracks without compromising
channel separation too much.
General purpose tape recorders use 2 mm tracks on a 6.3 mm tape.
On this machine the two tracks can each contain a separate mono program.
The tape recorder with 'stereo heads' uses a track width of 2.4 mm,
mainly at the expense of the space between the tracks and Cthus the channel separation.
Since the channel separation is of less importance for stereo material, this works fine.
As an extra bonus the track width offers about 1.6 dB extra signal to noise.
The butterfly heads write even wider tracks: 2.7 mm, adding another dB to the signal to
noise while the special head shape does prevent channel separation to degrade too much.
There is another problem with tape heads, called head bump.
I already mentioned that low frequencies cause a wider magnetic field.
This is not only perpendicular to the tape but also in the direction the tape moves.
That can cause a nasty resonance in the lows that depends on the tape speed.
In general the head design is made so that the resonance occurs between 40 and 50 Hz
at the preferred tape speed.
Since we talk professional tape recorders here, the preferred speed will often be 38
centimetres per second, 15 inches per second in the Queen's measure.
Tape speed can be chosen higher and lower with a factor two.
This will also change the head bump up or down with a factor 2.
There were artists that liked recording at 76 centimetre per second, 30 ips.
That would also shifts the bandwidth with a factor of 2, meaning that in theory a bandwidth
of 30 to 22.000 Hz would become 60 to 44,000.
Often the the playback head and the electronics limited the high end but more important the
head bmp would go from -say - 45 Hz to 90 Hz.
That bump could easily amplify the resonance frequency with 5 dB's on poorer performing
machines and still be 2 dB's in first class machines.
Since it is a resonance, the phase response is poor too.
That is why by nature analogue tape recordings can never sound as solid in the lows as digital recordings.
This makes it also interesting to see what MQA can solve here.
There is another reason why digital recordings will sound deeper and better controlled in the lows.
The playback head of an analogue tape recorder is almost the same as the recording head,
only the head gap is smaller since it need not to output magnetic force but rather read it.
Given the way the magnetic registration works, the output of the head doubles each octave
up.
Since we want an output that has an equal output for each frequency, an inverse filter is applied.
Here tape recorder manufacturers have to decide what compromise to make between the lowest
frequency and the signal to noise ratio.
If the filter is applied at 40 Hz, 20 Hz will be 6 dB's down but the signal to noise will
be 6 dB's better than when 20 Hz is chosen, as you can see in this graph.
Do realise that about 60 dB signal to noise was common on tape recorders
if you want clean recordings.
Pop and rock could be modulated higher since - in contrast to digital recorders - analogue
recorders have a very pleasant way of distorting when driven hard.
But Pop and rock didn't need the dynamic range and even it it did, about 70 dB would be the limit.
So choosing 20 Hz as filter frequency at the expense of 6 dB signal to noise was a bad
choice since no speakers then and only a limited number of speakers today would be able to
reproduce below 40 Hz.
As soon as half the wavelength of the signal on the tape reaches the gap width of the
playback head, the output will rapidly drop.
It might be clear that optimally the gap of the playback head is chosen so that the 20
kHz bandwidth defined by the DIN institute as upper frequency for hifi could easily be
played at the highest tape speed on the machine.
The better studio machines I have measured over time often got slightly over 20 kHz.
This means that when you digitise a tape at a higher sampling rate, you will still see
a rapid dropping of the output above 20 to 22 kHz.
Tape recorders can sound amazing, impressive and emerging because of qualities that were
and perhaps are still hard to achieve on digital media at 44.1 or 48 kHz.
Not for the limiting frequency range since a tape recorder doesn't do a better job.
But for the reasons I mention in my video "The truth about Nyquist and why 192 kHz does make sense."
In short: we are not able to band limit a signal to 20 kHz - as needed for 44,1 kHz sampling -
without causing severe time smearing that is most noticeable is the mid range,
something an analogue tape recorder does a good job in.
Tape recorders can't produce ultra deep lows and have a limited signal to noise ratio.
There also can be a heap of other problems like modulation noise - that's noise that
varies with the signal -, wow & flutter and all kinds of mismatching due to standards.
As with all analogue storage and transmission equipment,
the signal is 'shaped' to fit inside the medium.
In almost all cases the high frequencies are boosted prior to the medium to be attenuated
with the same amount when playing out the signal.
This also reduces the noise the medium had added and is therefore sometimes called passive noise reduction.
For tape recorders this principle is also used.
Unfortunately there are both consumer and pro standards and and of each
there is a European and a US version. In Europe we have the IEC setting the norm,
for the professional part driven by the European Broadcasting Union, the EBU.
I forgot who was responsible for the consumer standards in The States, but the professional
standards were set by the North American Broadcasters, the NAB.
A tape recorded with the IEC equalisation will not sound correct when played back on
a machine using NAB eq's and vice versa.
Playing back a tape with the correct EQ but the wrong tape head - say a tape with 2 mm tracks
played back on a machine with 2.7 mmpoles - will produce a faulty frequency curve in the lows.
Then there are tapes at several tape speeds, types of spools, wound tails-out or not and so on.
The tails out is yet another nice Granddad story.
If you load a reel of tape tape by putting it on the left teller to be wound on to the
right spool, you make your recording, spool the tape back on the left real and store it,
chances are that after a few years you will hear a faint echo in front of the music
- indeed a pre-echo.
This was caused by the transfer of magnetic load from one layer on the tape to the next.
It was a real problem until someone found out that when you didn't spool back the
tape after recording but stored it on the right spool - so with the tail out -
the print through would be after the original and thushard to hear
and when heard more natural than a pre-echo.
Didn't we have pre-echo's with digital as well?
Yes, but they are not audible as separate echoes.
I hope you found this interesting.
Please let me know either way so I can anticipate on what to discuss in my next Patreon Special video.
For I do appreciate your support immensely and really want to give you some extras.
Again, I hope you have enjoyed it and I hope you will enjoy listening to music even more.
See you in the next video.
No comments:
Post a Comment