Re: Sound card for recording

The message <slrnhhqivf.fjb.h@xxxxxxxxxxx>
from Tony Houghton <h@xxxxxxxxxxx> contains these words:

I'd like to make some recordings, maybe transferring some gems from my
vinyl collection to digital etc. I'm wondering whether I should use a
better sound card than the onboard sound.

I know one of the regulars here occasionally posts about an almost
universal design flaw on sound cards where the input circuitry clips
well below the potential ability of the A/D convertor or something.

That might have been me. It seems that _all_ the PCI soundcard
manufacturers "To A Man" (including the on-board sound used in
motherboards - all effectively connected via the PCI bus) just simply
used the "reference" design as issued by the soundchip manufacturers
without considering the implications of the reduced 5v unipolar bias
rail on the clipping limits of the line-in buffer amp.

The chip had a 'jumper link' option to desensitise the ADC input by 6db
(not a problem with a buffer amp powered by a 12v rail) which, rather
unfortunately, had been shown on the reference circuit as 'enabled'.[1]

I did try recording from vinyl via a T/T preamp to a basic onboard sound
card a few years ago. The preamp has no gain control and ISTR the result
did sound slightly distorted as if clipped.

How severe this effect sounds rather depends on how much headroom
you've allowed. The input clipping level corresponds to about -4db FSD.
If you were aiming to set the peaks to about 3db (thinking you'd still
have a 3db margin), it would only be obvious during the loudest
passages.

Professional kit seems rather overkill just to digitise a vynil
collection unless you can get hold of second user kit 'for a song'. If
you have the space, a cheaper alternative would be to use an older PC
that still has at least one ISA slot and fit a SB16 or AWE32 or AWE64
soundcard.

The AWE64 Gold has line input clip level around the +14db mark which
closely matches the line out clipping level of the RIAA pre-amp I built
into my record deck ;-) . The AWE32's clip level is about 3 or 4 db down
(similar to the SB16, afaicr) on this, but still well above the ADC
input clipping level.

The latching clip level indicators in Cool Edit Pro can usefully
indicate whether your initial attempt to digitise an analogue source had
produced FSD values which you can then home in on to ascertain whether
they were significant enough to warrant a second attempt.

Unfortunately, PCI soundcards totally ruin this feature, requiring you
to deliberately under-record by a good 10db and then check that
normalising to FSD required no less than a 5db boost to achieve (less
than 4db implying the peaks had been clipped before they hit the DAC
stage).

I've got one of those old Santa Cruz/Sonic Fury cards lying around,
which were reckoned to be pretty good in their time. Would it be worth
installing that for recording? I found an old review:

If that's a PCI card, experiment and check whether it bucks the trend
(unlikely, but still worth checking).

Audio Converters: Dual AC-97 2.1 audio codecs with hardware full-duplex
for simultaneous record and playback and up to 48 kHz sample rates.
18-bit A/D converters for high-resolution recording. 20-bit D/A
converters for high-resolution playback of up to 6 independent streams.

All that "digital bling" will be for naught if the line in buffer clips
4db below the ADC's line in level.

Alternatively I could probably borrow my friend's EMI external USB sound
adapter which he bought especially for semi-professional recording a few
years back. Significantly better still?

Maybe, assuming it really _is_ made to a 'semi-professional' standard.
I suspect it might be even more prone to line input clipping if it is
line powered from the USB socket on account of the reduced voltage
(typically 5v minus the reverse polarity protection diode volt drop of
0.3 or 0.7 volts, depending on the diode type used).

If the unit has its own seperate power source, this shouldn't be an
issue if the buffer amp is powered from a sufficently high enough
voltage to avoid clipping below the ADC's line in clipping level.

[1] I think this was achieved by programming the ADC reference voltage
from 2.56v to 5.12v (effectively, a peak to peak limit of the analogue
input voltage range). Reducing the reference even further just
'amplifies' the analogue input noise even further.

The choice of 5.12v was obviously a noise issue based decision without
regard to the effect of using a 5 volt rail in place of the 12v rail
used by the older ISA based designs (which could even have had the
benefit of a dual rail supply <+/-12v>).

Given a sensible choice of ADC reference voltage, even a decent USB
dongle style of converter could do better than the vast majority of PCI
based soundcard designs in this regard just as long as you have a means
of attenuating the line in level coming from the analogue source.

--
Regards, John.

Please remove the "ohggcyht" before replying.
The address has been munged to reject Spam-bots.

.

Relevant Pages

  • Re: A reasonable argument against double blind tests?
    ... proclamations on the electrical design of components, and I certainly never said it did. ... However, one does not need to be an "authority on electronics" to see the silliness we have with certain attributes of high-end, "serious" audio. ... However, many of those amps are much more expensive than necessary to do the job, and I think it is the duty of journalists to point out that fact to potentially gullible readers. ... certainly a concerned audio buff would be wise to take some time to learn a bit about the sound of components. ...
    (rec.audio.opinion)
  • Re: potato-potato
    ... But, they do sound sweet. ... he hears the crisp exhaust note of a ... Harley engine--a sound one Honda spokesman characterized as, no kidding, ... their design engineers are asked to create products that are technologically ...
    (rec.motorcycles.harley)
  • Re: tv receivers have puny sound
    ... I used to design marine radios. ... Thin sound means no bass. ... Good speakers require a rather solidly built enclosure. ... In locations where the radio volume was run at full blast to ...
    (sci.electronics.repair)
  • Re: It is my birthday and YOU get the present
    ... Abbeddrose Bierce wrote: ... If you had a clue as to what is necessary to design the most ... When you say "instruments" you mean the MPs, ... to get the sound in his head. ...
    (rec.music.classical.recordings)
  • Re: tired...waiting for GSOUND
    ... ATMega32, SD card, PWM sound amplified by a LM386 and a few ... sure not HiFi but way better than any original early sound card. ... directly to the driver transistors or the flip-flops. ... There's usually a solenoid for resetting ...
    (rec.games.pinball)