locked Vertical spikes every 1000 hz prevent decoding


 

USB has a synchronising pattern of bits every 1 ms. Also every 125 us for higher speed USB. I had this problem with one USB cable and I replaced it with a shorter shielded cable.

 

73 Phil GM3ZZA.

 

Sent from Mail for Windows 10

 

From: pa3abk
Sent: 15 July 2021 17:13
To: main@WSJTX.groups.io
Subject: Re: [WSJTX] Vertical spikes every 1000 hz prevent decoding

 

Thanks OP for starting this... I was struggling with this problem with a cheap USB soundcard from Ali Express.

Type 5HV2 it has a ferrite in it's lead and this should be a warning when buying... it has an interference issue.
Inside the chip, is completely covered with a substance which is difficult to remove. If you compare the layout with the manufacturer suggested  you will miss several interference reject elements like ferrite beads.
I tried to silence the unit as outlined in one of the messages, but there is still a "loop" giving the 1kHz interference. Possible the return minus voltage must be separated ...

In AUDACITY you can make a frequency analyse which confirms this product is completely useless for our use. (For gaming it will work.)

Normally I use an UCA222 from Behringer which performs excellent! The noise level on that device is -110dB!

Jan/pa3abk

 


--
73 Phil GM3ZZA


bobo7493@...
 
Edited

Hi Jan,

i had also found a good program for testing the noise Level on USB Soundcards "DL4YHF Audio Spectrum Analyzer" 

https://www.qsl.net/dl4yhf/spectra1.html

There also some noise tests from some cheap usb soundcards on this side - tested with this program and how to use it for that...

https://dl1gkk.com/sound-card-check/

Hans / DB5ZP


Michael Black
 

You can't measure noise floor this way.  You have no idea what level those sound devices are representing as none of them are calibrated except your rig which is still doubtful.

You need a reference signal so you can calibrate each device to a known source before you measure noise floor.  So say a -70dB signal shows -70dB on the spectrum (assuming high resolution spectrum -- if it's a curve the power is integrated under the peak curve).  You need at least two reference points to ensure linearity too.  So measure -70dB then check below that and above it to see if it's linear.

I can do that here with a 2 Signalinks and a couple of rigs and a signal generator.  I'll do that and post the results using the same Spectrum Lab settings.

Mike W9MDB






On Friday, July 16, 2021, 06:23:45 AM CDT, bobo7493 via groups.io <bobo7493@...> wrote:


[Edited Message Follows]

Hi Jan,

i had also found a good program for testing the noise Level on USB Soundcards "DL4YHF Audio Spectrum Analyzer" 

https://www.qsl.net/dl4yhf/spectra1.html

There also some noise tests from some cheap usb soundcards on this side - tested with this program and how to use it for that...




Michael Black
 

I was unable to measure the noise floor on my calibrated ANAN 7000DKE MKII -- it runs around -143dB and that's just too small to register on the ADC.  I can increase the gain but that is a bogus measurement.  Increasing the gain the lowest level SpectrumLab shows is around -130dB.

So it would appear to me that SpectrumLab is unable to measure the real noise floor.

Mike W9MDB


On Friday, July 16, 2021, 05:56:10 AM CDT, bobo7493 via groups.io <bobo7493@...> wrote:


Hi Jan,

i had also found a good program for testing the noise Level on USB Soundcards , there also some noise tests from some cheap usb soundcards on this side :

https://dl1gkk.com/sound-card-check/

Hans / DB5ZP




Alan G4ZFQ
 

I was unable to measure the noise floor on my calibrated ANAN 7000DKE MKII -- it runs around -143dB and that's just too small to register on the ADC.  I can increase the gain but that is a bogus measurement. Increasing the gain the lowest level SpectrumLab shows is around -130dB.
So it would appear to me that SpectrumLab is unable to measure the real noise floor.
Mike,

Seems to me there are unknowns. Soundcards have different ADCs, gains, input noise.
Essentially standards are required.
I'm no expert but I'd attempt to test the SNR of a weakish standard. It may be necessary to play with gains. Then the resulting dynamic range is important.
SpectrumLab is as good a tool to do this as any other but I don't think a single check on different cards without a standard will prove anything.

73 Alan G4ZFQ.