пятница, 15 мая 2009 г.

Record for the following analysis.

If you record the signal just for yourself, then you can do not read all that follows bellow. But if you are making
the record, specially, to represent it to public, or to give it for further analysis, then you must take care about
quality of this record.

Good quality of the record of the signal – is the half of success.

1. Never record with digit capacity of 8 bits. This 8bit mode suits to everything, but only not for finding-out of
nuances of the signal’s structure. Use only 16bit. That is not the point, that 8bit it is bad, the point is that
16bit, by default, provides the quality, which is enough for analysis and disassembling, while 8bit provides such
quality only in very rare cases. There is also no sense to convert the record from 8bit into 16bit artificially – it will be self-deception: all subtle connections inside the signal will be lost irretrievably.
Forget about recording from a microphone, it will be the fatal end of your record, don’t even not think to publish
it, only a linear output, or in extreme case, an output on ear-phones.

2. Never write the record in the compressed format at once. The record if it is really interesting to you, and in
some sense is precious, should be written in the pure form, without a compression – in the old kind WAV format.
underline, not some sort of WAV format(for example ADPCM and many others), but namely the pure PCM WAV standard. The original record should have the highest quality, then, if possible, it can be compressed, but if to compress it at once - an initial clear picture of the signal will be lost and not available for restoring any more. This is very bad because, there are lost of situations, when not compressed signal can be demanded easily.

3. It is necessary to approach with all gravity and seriousness to the record’s sampling rate - for a large number of signals the sampling rate should be not below then 11025 Hz, that provides the upper boundary of a range on the level about 5.5 kHz. For the record of satellite signals it should not be lower then 22050 Hz.
Do not hope that it is enough 8kHz sampling rate for telephone channel, all this is true, but in terms of monitoring,
particularly on HF, the inaccuracy of the settings and features of the modulation can easily deflect the signal from
the telephone channel and the "unnecessary" half kilohertz is not only interference, but the guarantee of deflection.
Only after the record is done, it is possible to reduce the sampling rate, by results, if it is required. The records
of satellites are, generally, easily overlapping upper boundary for 6-8 kHz , thus, remember - the sampling rate for them should not be less than 22050 Hz!

4. Never make the record without preliminary check-up/setup(setting), of path(tract,route) of the record. There is nothing worse, than the rare record, which is made with an overload.
Such records do not cause anything but regrets, it is impossible to understand them, to result them into normal kind is also impossible, you can only throw them out into basket.
Pair of the minutes, which are have been saved by the lack of knowledge or by the laziness, will expressed in a waste of time and an absolutely useless resources.
The maximum signal’s level at the recording should not exceed half of the possible maximum, thus guaranteed the
existence of at least some reserve, in case of any surprises.
It is better to write with certainly reduced level, than with certainly high.
Weak level can be raised, while the overload in the record cannot be removed by any weakening , the signal’s
form/shape will be distorted to unrecognizability.

5. If you turn the receiver with purpose to write something "special", then immediately turn on the record. There are
a lot programs, which are writing the record into long file, personally I use RecAll, it writes, while I’m busy withmonitoring. The point is that the rare and interesting moments are always occur unexpectedly, and they are very short, the vanity, which is arising from convulsive attempts to launch /run/start recording program, leads to the fact, that neither the record nor the signal is there in the end. When the record goes permanently, the risk to skip something interesting is about zero, it is necessary only to be provided with reasonable time of the record into "a ring" file, for me this reasonable time is about an hour.

6. Do not do anything with the record, do not improve it and do not try to make it look better, leave it fully "as it
is". You can easily corrupt it, and what you seem as " unnecessary" can play the very important role in the analysis.

7. Okey, so the record is have written, and everything seems good to you, you want to public this record for an
identification or for other purposes – please be careful, do not create illusions, view the record by any editor or a
software, and check what is exactly you got in that record.At least estimate the form of the signal: there should not be any restrictions by the amplitude or any unexplained strangenesses. If you do not understand what is happened within your record, then others, most probably, won’t want to understand it especially.

And in conclusion: To make a good record - it is same work and skill as all other professional skills, nothing can
be well-done by itself. Train, try, play in quiet conditions. Look what is overload on practice, check what reasons
cause overload, try to remove it, learn how to define it.
It is not enough just to have a knowledge, it is necessary to know how to apply this knowledge, the only one method to know it – is the practice and experience.

Author SergUA6