On 17/02/2021 11:08, Reino Talarmo wrote:
*>From:*Bill Somerville [mailto:[email protected]]
*Sent:* 17. helmikuuta 2021 0:16
*To:* [email protected]
*Subject:* Re: [wsjt-devel] Clipping
On 16/02/2021 22:10, Jim Brown wrote:
On 2/16/2021 2:01 PM, Bill Somerville wrote:
there are no analog audio stages involved. The trival
distortion (lower than -65dB) is probably the result of being
passed through an unnecessary sample rate converter with a
small defect within the MS Windows audio sub-system. The issue
can be demonstrated without ever leaving the digital domain.
I understand that, but I understood Michael to be saying that he
used one computer to look at the audio output of the other. Yes,
-65dB is pretty low, trivial unless the guy transmitting it is 40
dB over S9 and you're trying to work weak signals. :) Not an
uncommon situation in Northern California on 160 and 6M.
73, Jim K9YC
Hi Jim,
I believe you have misread what Mike did, he used two instances of
WSJT-X on the **same** machine using a digital only loopback like VAC
or VB Cable.
73
Bill
G4WJS.
Hi Jim and Bill,
There is still one possible mechanism to be checked. If I have
correctly understood what Mike did i.e. he used Tune to generate about
1500 Hz tone and fed that to the WSJT-X receiver. All were almost
perfect below a tone level and above it bad looking “sidebands” at
many frequencies was seen on waterfall. I tried those files
210214_211315.wav and 21014_211345.wav and saw the same strong looking
artifacts. The essential issue was that those artifacts were strong in
the waterfall, but weak at spectrum display. More importantly the
artifacts were stong only, when I used “Low sidelopes” and “Flatten”
activated. Some minor ones are present (+/- 300 Hz), when I deactivate
Flatten. With “Most sensitive” I cannot see any artifacts.
How flattening is performed may affect to the result, but filtering
seems to affect more.
In any case my conclusion is that those artifacts are due to receiver
signal processing and so not really transmitted at all. I have not
tried to analyze those files using some independent spectrum analyzer.
73, Reino OH3mA
Hi Reino,
your conclusions are incorrect.
We have verified that WSJT-X produces clean audio when the Pwr slider is
at 0dB. This has been done by analysing looped back audio on Linux and
macOS as well as by temporarily amending WSJT-X to write a PCM .WAV file
with exactly the samples that are sent to the audio output device. In
all cases the data is exactly as expected, i.e. single tone without
harmonics, sidebands, or any other distortion.
I have also verified that the same audio when recorded via a loopback
like VAC or VB Cable on Windows has the distortion added, the distortion
is not harmonics, it looks like close in sidebands but not a regular
"comb" appearance. I suspect it due to some internal audio processing
within Windows (perhaps a defective re-sampler algorithm that is always
enabled), but it is not possible to rule out VAC or VB Cable although
the fact that they produce identical results does tend to imply the
issue may lie within the Windows audio sub-system.
The distortion components are more that 65 dB down on the main tone so
are trivial and below the likely distortion output of almost any
transmitter used by Amateur Radio operators. If you are really
concerned, although there is no need to be, then reduce the Pwr slider
by 0.5 dB and you will have no distortion. Alternatively, do not use MS
Windows ;)
73
Bill
G4WJS.
_______________________________________________
wsjt-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
