On Aug 26, 2013, at 11:22 AM, W5XB wrote:
> Chen, Joe, or others, has anyone tested/posted a list of sound card
> performance? They can't all be equal!
No, they certainly are not equal.
At the instigation of Steve Ford, I had started to write a white paper on sound
cards earlier this year, but tons of other things took precedence. I may one
day get back to finish it after my the top of my stack is cleared.
If you want some raw numbers that I had measured to write the article, here are
some of them that I measured.
They are for receive direction only, I did not measure the transmit direction.
I do suspect that some of them have a high enough harmonic distortion that may
not be so nice when you transmit with wideband AFSK.
You may need to reformat the tabs in the following. I have no idea what it
will look like on your email reader, I have chosen Rich Text, so you may also
have to change your reader to that to get the correct tabbing -- if too many
people have problems with the tabs, I can spend a little time to reformat it
into a pdf file and post the URL. But lets try this simple posting first
though.
I have listed the 2nd harmonic distortion, the 3rd harmonic distortion, the
noise floor (NF) and third order IMD (3IMD). You subtract the noise floor from
the input level and you have an estimate of the dynamic range. The DR ranges
from 118 dB for the Presonus to about 96 dB for the microKeyer, to 98 dB for
the SignaLink USB, to 82 dB for the Griffin iMic, to 96 dB for the Amazon
special (that is where I bought the Syba from).
The Presonus is the best 24 bit sound card I own. The SignaLink USB uses the
ubiquitous PCM 2909 chipset, so you expect other sound cards that use that
chipset to measure the same in harmonics and IMD, but potentially differ in
noise floor.
If you are using a sound card with a narrow receive crystal filter, you can
ignore the harmonic columns. The IMD column is still important if some large
signal is leaking through the crystal filter.
By the way, IMD is really only important when you have two large signals of
comparable strengths. When one of them is even 10 dB less loud, all problems
go away (2dB drop for each 1 dB change for typical 3rd order IMD from
soundcards, transmitting PA, etc) -- this is why SDRs still work wonderfully in
a crowded band. There are just not enough signals near the clipping level. And
that is also why soundcards work quite wonderfully for wideband waterfalls and
almost never clips because of many moderately strong signals -- they usually
clip because of one single very strong station.)
The tones I used are 1100 Hz and 1200 Hz. The noise floor, if memory serves (I
will need to check the code to be sure) is from a 32K FFT, running at 48,000
samples per second. But for comparing purposes between the different sound
cards, you don't really need to know the absolute noise bandwidth.
None of the sound cards shows any gain compression up to full scale, that is
why the "input" and "generator" columns are the same. The "--" are noise
limited values (so are any reading that has a "n.l." next to them).
I have included in the measurements the peak voltage levels where the sound
card clips. Notice how different some of them are (the smallest and largest
full scale levels in the tables is a whopping 20 dB between the SignaLink
(relatively deaf) and the Amazon special (really meant as microphone input and
not line level)). This could account for the majority of problems people see.
Proper adjustment of a sound card is *crucial* to good weak signal copy.
The picture that I took for the (vaporware) article is here, it shows the sound
cards that I measured:
http://www.w7ay.net/site/Downloads/soundcards.jpg
Others can send in and share what they measured.
These ones are just a small set of the sound cards that I use with my radios,
and some, like the SignaLink USB and $8 "Amazon special" (Syba UAUD), purchased
just to make measurements for the article. I also have the E-MU 0404 and
Roland FA-66 and a myriad of M-Audio and TASCAM soundcards, four generations of
Griffin iMics (all iMics have different performance; the current version
(white) is the worst, the old gray ones were much better, and were the ones
that made a name for the iMics).
For the (vaporware) article, I thought the following five are representative of
both performance and price range.
The "$8 Amazon special" is surprisingly good, but its problem is that it is
unstable for some generator impedances. This is what I had written in the
vaporware article:
> The Syba SD-CM-UAUD is a sound card that incorporates the ubiquitous 16-bit
> CMedia CM119 (Taiwan) audio codec. There are many similar CM119 based sound
> cards on the market, and they can be purchased for anywhere between $2 to $8
> each.
>
> The CM119 based sound cards are easily identified by having a monophonic
> microphone input. All the other interfaces that are tested in this report
> support stereo input; the Presonus supports 8 inputs.
>
> Physically, the Syba "enclosure" includes just two 3.5 mm jacks for analog
> input and output, together with a USB Type A plug for connecting to the
> computer. The sound card is bus powered. It is shown at front center of the
> title photo.
>
> In addition to having no external controls, the input only supports two
> sampling rates of 44,100 samples per second and 48000 samples per second.
> Further, the input codec shares the sampling rate with the output codec. If
> you select 48000 samples per second for input, you must also use 48000
> samples per second for output.
>
> This particular sound card appears to be only conditionally stable when
> driven by a high impedance source. It appears to be stable when driven by a
> source resistance of 91 ohms or less. I do not know if this is common with
> all CM119 based cards – if you experience any sudden burst of poor RTTY
> decoding, you should check to make sure the radio's output impedance is
> sufficiently low to drive this device.
>
> The CM119 appears in the Mac OS X USB Device Tree as a Generic USB Audio
> Device. The device has a software controllable attenuator with a range of
> about 23 dB. In the least sensitive setting of the attenuator, the Syba clips
> with a sine wave of 112 mV RMS. In the most sensitive setting of the
> attenuator, the Syba clips with a sine wave of approximately 8 mV RMS.
>
> The input of this sound card is obviously built for low level microphone
> inputs. The output of receivers with fixed line level output will most
> definitely need to feed this sound card through a resistor attenuator.
>
> You should avoid this sound card if you are not prepared to provide it with a
> low voltage drive from a low impedance source.
Anyhow, the tables follow.
73
Chen, W7AY
=========================================================
Presonus FireStudio Mobile (Master Level -0.2 dB for full scale == 242 mV)
Generator Input 2nd Harmonic 3rd Harmonic NF
3IMD
0 dB 0 dB -90.4 -76.8 -117.5
-2 -2 -95.5 -82.2 -119.2
-4 -4 -101.7 -87.7 -120.9
-6 -6 -106.8 -93.0 -122.7
-86.7
-8 -8 -111.1 -98.0 -124.1
-89.5
-10 -10 -114.9 -101.9 -125.5
-92.3
-12 -12 -116.3 -105.3 -126.6
-95.3
-14 -14 -121.3 -108.5 -127.4
-98.3
-16 -16 -126.7 -110.8 -128.2
-100.9
-18 -18 -130.5 -113.2 -128.5
-103.6
-20 -20 -134.5 -115.6 -128.8
-106.4
-22 -22 -135.8 -117.2 -128.8
-109.2
-24 -24 -- -119.7
-128.8 -111.6
-26 -26 -- -122.1
-128.9 -114.0
-28 -28 -- -124.9
-129.0 -116.7
-30 -30 -- -128.1
-129.0 -119.5
==========================================================
microKeyer II (Master Level -0.6 dB for full scale == 316 mV)
Generator Input 2nd Harmonic 3rd Harmonic NF
3IMD
0 dB 0 dB -84.4 -65.3 -96.1
-2 -2 -90.6 -69.4 -97.7
-4 -4 -95.1 -73.4 -97.7
-6 -6 -99.5 -77.7 -97.7
-65.2
-8 -8 -103.4 -83.4 -97.7
-68.5
-10 -10 -106.5 -87.0 -97.7
-72.2
-12 -12 -108.0 -89.1 -97.8
-75.9
-14 -14 -- -90.9
-97.8 -79.3
-16 -16 -- -93.1
-97.8 -83.4
-18 -18 -- -96.4
-97.8 -87.3
-20 -20 -- -99.1
-97.8 -91.4
-22 -22 -- -102.6
-97.8 -95.4
-24 -24 -- -105.2
-97.8 -99.5
-26 -26 -- -107.8
-97.8 -103.0
-28 -28 -- --
-97.8 -106.3
-30 -30 -- --
-97.8 --
==========================================================
SignaLink USB (Master Level -0.2 dB for full scale == 1125 mV)
Generator Input 2nd Harmonic 3rd Harmonic NF
3IMD
0 dB 0 dB -102 -59.3
-98.2
-2 -2 -98.2 -61.0 -98.4
-4 -4 -93.6 -62.6 -98.7
-6 -6 -98.4 -64.1 -98.8
-60.1
-8 -8 -95.1 -65.6 -99.0
-62.1
-10 -10 -95.2 -66.9 -98.9
-64.3
-12 -12 -98.1 -67.7 -98.8
-66.3
-14 -14 -102.8 -69.5 -98.0
-68.2
-16 -16 -101.0 -70.9 -98.5
-69.8
-18 -18 -104.0 -72.5 -98.1
-71.2
-20 -20 -104.0 -74.2 -98.8
-73.1
-22 -22 -104.0 -75.1 -99.0
-75.2
-24 -24 -104.2 -76.8 -99.0
-76.7
-26 -26 -105.5 -78.6 -99.0
-78.2
-28 -28 -- -80.3
-98.8 -79.9
-30 -30 -107.0 -82.1 -98.7
-81.8
==========================================================
Griffin iMic (Master Level -1.02 dB for full scale == 405 mV)
Note: 1 KHz spurs at -80 dBc.
Generator Input 2nd Harmonic 3rd Harmonic NF
3IMD
0 dB 0 dB -73.7 -73.3 -82.0
-2 -2 -77.1 -93.0 -84.9
-4 -4 -78.8 -95.8 (NL)
-84.9
-6 -6 -80.5 --
-84.7 -89.2
-8 -8 -82.1 --
-84.7 -92.2 (NL)
-10 -10 -83.2 --
-84.7 --
-12 -12 -84.1 --
-84.7 --
-14 -14 -86.4 --
-84.7 --
-16 -16 -88.3 --
-84.7 --
-18 -18 -91.2 --
-84.7 --
-20 -20 -92.4 --
-84.7 --
=======================================================================================
Syba UAUD (C-Media CM119) (Master Level -0.5 dB for full scale == 112 mV)
Note: Unstable if high impedance drive; use 91 ohms
High order IMD
Generator Input 2nd Harmonic 3rd Harmonic NF
3IMD
0 dB 0 dB -76.6 -80.9 -96.1
-2 -2 -82.0 -82.1 -96.4
-4 -4 -84.9 -87.4 -96.2
-6 -6 -81.1 -105 (NL)
-96.2 -88.7
-8 -8 -81.8 -99.0 -96.4
-91.8
-10 -10 -84.9 -104.0 -96.3
-95.0
-12 -12 -88.5 -108.4 NL
-96.5 NL
-14 -14 -92.4 --
-96.3 --
-16 -16 -96.1 --
-96.3 --
-18 -18 -99.7 --
-96.4 --
-20 -20 -102.6 --
-96.5 --
-22 -22 -- --
-96.5 --
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|
