Re: LF: Re: BBB-4-like receiver ready for first tests

[Daniele Tincani <danieletincani@yahoo.com>]

Hello Alan, LF, VLF,

 

thank you very much for your suggestions about how to perform more sensible measurements.

 

The Agilent 33120A is a general-purpose function generator. Among other functions, it is specified to produce a noise signal (gaussian) with 10MHz bandwidth. Amplitude can be set between 50mVpp and 10Vpp into 50ohms, on open circuit (probably on the BBB-4 hi-Z input) the amplitude levels become 100mVpp to 20Vpp.

 

At 50mVpp (I admit: with no particular precautions about case leakage, coupling, etc.), it looks pretty flat on SpecLab at about -114dB (long-term average value), at least up to 22KHz with 48KHz sample rate.

 

The souncard alone shows a noise level which is much lower. Please see my attachments.

Best regards

Daniele

---

From: Alan Melia <alan.melia@btinternet.com>
To: rsgb_lf_group@blacksheep.org
Sent: Tue, November 9, 2010 2:33:36 AM
Subject: Re: LF: Re: BBB-4-like receiver ready for first tests

I worry that Daniele may be confusing himself here. I dont know that noise
source but is it specified for producing audio white noise?? If not then it
cannot be relied upon to be flat at audio. I made a noise source with a
zenner diode an a couple of transistor stages like the standard noise bridge
source, but bigger coupling caps. This gave a genuine flat response on spec
lab or Argo from low frequencies to 48kHz (96ksampling Delta 44)
If the agilent is intended to work on a 50ohm system then the level will be
approximately twice that on the scales. Most audio generators would be
expected to operate around 600 ohms. However there is another problem case
leakage makes most audio generators unreliable for "radio" type work below
1mVrms. The solution maybe to move the generator away from the receiver put
an potential divider direct on the receiver input and use screened balanced
output preferably with a centre tapped ground from the generator, not
connected at the receiver.

I would not have thought a whistler receiver should have a peak response at
2kHz?? Most whistler plots I've seen show best responses 4 to 9kHz.

The BBB-4 seems to have a filter in it and I think it would be best to
measure the response with a calibrated audio generator with a termination
resistor across the BBB-4 input. Paul Nicholson has done a lot work and
published the result here on the noise in various front end configurations.
Also measuring low levels of noise with a PC is fraught with difficulty
because they produce so much themselves. You must be very careful that you
are not measuring PC noise. I prefere an oscilloscope or even an audio
mVmeter, You must also remember there may be unexpected effects from you
modification. That  headphone lead in use, with the extra gain, can act as a
transmitting aerial!!  So you really need to use screened twin cables here.

Alan G3NYK

----- Original Message -----
From: "Stefan Schäfer" <schaefer@iup.uni-heidelberg.de>
To: <rsgb_lf_group@blacksheep.org>
Sent: Monday, November 08, 2010 6:10 PM
Subject: Re: LF: Re: BBB-4-like receiver ready for first tests


> Hi Daniele, Jim, LF,
>
> Am 08.11.2010 16:47, schrieb Daniele Tincani:
> >
> > Also consider that when I created a short circuit on the antenna input
> > of the rx (see Q2 in my e-mail), I got a response curve on SpecLab
> > similar to that I had with the Agilent, but with a peak level about
> > 30dB lower (about -107dB around f=2KHz on SpecLab).
> This means that your generated noise level was to low and you have
> observed the soundcards noise, at least outside the region of 2 kHz. So
> just look at 9 kHz without the generator and then increase the noise
> gernerator level so that the noise increases say 20 dB.
> > ------------------------------------------------------------------------
> > *From:* James Moritz <james.moritz@btopenworld.com>
> > *To:* rsgb_lf_group@blacksheep.org
> > *Sent:* Mon, November 8, 2010 3:22:59 PM
> > *Subject:* LF: Re: BBB-4-like receiver ready for first tests
> >
> >
> > The filter rolls the gain off rapidly below about 1kHz and above about
> > 10kHz. So this would be OK for whistlers and 9kHz reception, but would
> > attenuate VLF utilities at higher frequencies.
> >
> Normally the MSKs are so strong that an attenuation of say 20 dB
> (compared to 9 kHz) will probably no problem. 20 dB gain reduction would
> not even mean 20 dB S/N reduction...
>
> What i would worry about is the high gain arround 2 kHz, where the mains
> hum is dominant, even if the frequency response (without an input
> signal) would be flat. Maybe this could become a problem if the levels
> get so high that the amp stages become nonlinear and/or the soundcard
input.
>
> But just try what happend in reality!! In my experience it is a good
> indicator to see the diurnal noise levels having a minimum at arround
> 8...10 UTC. If the level difference is about 10...15 dB (depending on
> the WX of course) it is a good first step to assume a sensitive RX. The
> rest can be seen in tests where a far field signal is generated on the
> Dreamers Band ;-)
>
>
> 73, Stefan/DK7FC
>

Agilent_33120A_Noise_50mVpp.jpg
Description: JPEG image

Soundcard_Noise_24bits_48KHz.jpg
Description: JPEG image