Confused by soundcard measurements

[Christer]

I would like some help to think and understand what is wrong,
or if anything is wrong at all.

I am working on a program to measure RMS values via the PC
soundcard. I plan to add some more functionality, but right
now I am just trying to get the RMS measurements right.
They measurements confuse me, however. Currently I compute
the RMS value and the mean magnitude of the samples over
a 10 s. period of time. I sum the squares of the sample values,
as an approximation to the integral it should be if time weren't
discrete. This should give accurate enough results. However,
in addition I also keep track of the ten highest magnitude
values. I call the highest of these values the peak value.

Now to the confusion. If I measure a 1k sine from my CDP
I get very consistent results between measurements, the
peak values being almost identical between measurements.
However, the peak to RMS ratio is 1.22 rather than 1.41 as
it should be and the RMS to mean ratio is 1.07 rather than
1.11 as it should be. That is, I seem to get too high RMS
values (the level is low enough to avoid clipping). If i instead
use loopback with the soundcard and use the same wave
file as is recorded on the CD I get a different result. The
peak values differ quite a lot between measurements, so
the peak to RMS ratio varies from 1.41 up to over 1.6
between measurements. The RMS to mean ratio is 1.11 as
it should be, however, and this is consistent between
measurements. Using loopback and the signal generator
from Dazyweb Labs I get consistent readings between
measurements, but a somewhat too high peak to RMS ratio
of 1.47.

I have wrestled with this problem and I find it hard to see
it could be caused by some bug in my program. A bug ought
to cause a more systematic error since the program doesn't
know what it is measuring. Does anybddy have any ideas
what is causing these strange results?

The above results were taken with a 24-bit soundcard in
16-bit mode. I have also used an older 16-bit card and
get similarly strange results, but not quite consistent with
those above.

[hjelm]

Only guessing here but could it be noise?
Do you have any figures on signal level and SNR at those levels.

Another thing are you sure you are not measuring a ground loop or something. Do you have a visual of the signals?

[Christer]

Originally posted by hjelm
Only guessing here but could it be noise?
Do you have any figures on signal level and SNR at those levels.

Another thing are you sure you are not measuring a ground loop or something. Do you have a visual of the signals?

I had been thinking of noise, but the errors were too big and
not consistent with noise. Neither is it consistent with ground
loops and I can't see there could be one when running
loopback. However, while I had looked at the signal as recorded
wave files (my program has no visualization) I hadn't done so
after swapping to the new soundcard. I decided to check again
after you post and it seems there were two independent
problems, which added to my confusion. The new soundcard
is obviously not healthy. If I run loopback and record what I
get as wave files, the signal is heavily corrupted with strange
bursts. There is no such issue with the old card, which also
was consistent for both loopback tests when checking that
again now. I still seem to get somewhat strange results, but
not as much as with the new card, so maybe it can be explained
by noise or sampling issues now? At least one error source
is removed now and will go back to the shop it came from.

[peranders]

Christer, have you downloaded RightMark Audio Analyzer? Use this software to check yours with.

Russian freeware
http://audio.rightmark.org/

[Christer]

Originally posted by peranders
Christer, have you downloaded RightMark Audio Analyzer? Use this software to check yours with.

Russian freeware
http://audio.rightmark.org/

I know it is more or less standard for soundcard tests but I
have tried it now and things are messier than they ever were
now. I don't think I've seen such a buggy and crash-prone
program in years. It crashed really bad several times and
seems to leak resources or something.
I can't get it to work at all on my old Soundblaster card.
I can't adjust the levels high enough in the
mixer. The strange thing is that the first time I tested, I could
at least set the levesl to around -7dB, which was still too low
to run the tests. After swapping soundcards to the new one
and then back, I cannot get the level above -70dB!!! And that
even after reinstalling Windows from a drive image and
reinstalling the Creative software, that is, all traces of the new
soundcard should be gone. On the new soundcard I could run
the test at first but then got problems when trying again
later. When in level adjust mode, I can see hints of abrupt
level changes, which may explain the strange bursts I saw
on recorded data earlier. Right now there are several things
that no longer works even after reinstalling windows. I am
starting to believe the lower PCI port has been taken over
by aliens. "All your soundcard are belong to us!"

I am more puzzled than ever right now. My own program
that worked although giving some strange results no
longer works at all, and I haven't even recompiled it!!
Progams don't change without being recompiled (at least
not if written in decent a decent language) so the problem
must be elsewhere. Two broken soundcards?? Sounds
improbable, and it is no problem to record and playback
music as usual. It is just any type of loopback test that fails.
And yes, I have measured the loopback cable, trying to
torture it meanwhile.

[johnnyx]

You could try uninstalling and then re installing the soundcard drivers, I had strange problems with Praxis because of this, the files somehow got corrupted.
Another thing is that the Windows OS can interfere with the timing of samples, so they may be jumbled up, so to speak. For example, Praxis uses one stereo channel for the signal, and the other for a timing pulse to synchronise the samples.
Either of these may be the cause of your problems, I hope it helps.

[Christer]

Originally posted by johnnyx
You could try uninstalling and then re installing the soundcard drivers, I had strange problems with Praxis because of this, the files somehow got corrupted.
Another thing is that the Windows OS can interfere with the timing of samples, so they may be jumbled up, so to speak. For example, Praxis uses one stereo channel for the signal, and the other for a timing pulse to synchronise the samples.
Either of these may be the cause of your problems, I hope it helps.

Thanks, but I had a drive image of C: where the up-to-date
drivers were installed, and after RMAA crashed so badly
that it even refused to run the level adjust I decided to
bring up a fresh Windows from this image. I had done it
yestaerday anyway, so basically lost nothing by doing it.
Maybe I should try to reinstall anyway. There is only one
thing one knows with windows and that is that one never
knows.

[Christer]

Hm, I reinstalled the drivers and now the left channels works
to some extent. The max level I can get in RMAA now fluctiates
between -15 and -10dB, which is still too low, and the right.
Something is definitely not well somewhere, and it is not getting
any better by RMAA constantly crashing.

[hjelm]

Do not know if it is anything real or just my imagining.
I have a laptop which seems to be very confused when i use the keyboard increase/decrease volume. It doesn't change the volume setting bar but the volume itself changes. The odd things occur when i try to adjust the volume bar then the volume jumps to the bar position and i can adjust it via the bar.
The most annoying thing though is when start some programs, explorer sometimes, or do some operations in some programs the volume jumps back to the volume bar level. It is like both ways affect the volume but only the keyboard takeds the previous setting of the soundcard into consideration, the otherone remembers its own previous setting.

[Christer]

You are probably not imagining things. It seems my problems
are similar, altough somewhat different. I am becoming more
and more convinced there are problems with the drivers and/or
windows, most probably with the combination of them. I think
Creative has a bad reputation for their drivers so maybe one
should expect strange things to happen. Now when I have
reinstalled drivers for both cards, the meters in Creatives
recorder indicate that the left channel is much louder, but only
at loud settings. Checking the recorded file shows no sign of
any imbalance, however. Trying to run RMAA on my old card
now results in the left channel working but at -7dB, which is
too low, and the right channel is dead. Yet, there is no problem
to record or playback music for instance. I wish there were
some cheap but decent soundcards without mixers and other
crap that can mess things up.

I bought my new soundcard because I suspected there was
something wrong with my old one, but I have it on a 30 day
trial period. Maybe I should hand it back and buy something
else. It seems people often prefer Terratec cards to Creative.
Are there any opinions on this, especially if Terratec has
better drivers?