Topics

locked Spyserver on Raspberry PI 3 (+) vs. Direct Connect by USB #spyserver #raspberrypi

matrixxx <dm5xx@...>
 

Hi airspy,

Some strange behaviour with a higher noise floor while the antenna is not connected:
- if i connect the airspy hf+ directly to my pc, i see about -135dbm noise floor (using sdrconsole)
- if i connect the airspy hf+ to my spyserver running on a raspberry pi 3 (and 3b+, tested both), the noise floor is about 20dbm higher at -115dbm, which is A LOT!

To eleminate the interference i:
- connect to the spyserver via Wifi network (so no ethernet cable)
- attached a battery to supply the the raspberry and the airspy hf+ both with 5v enough power
- put the raspberry completely into a metal case

Nothing helped... So it seems that the noise is coming out of the USB socket (and/or the usb cable is bad, with just 2 cables connected, one of them has to be the problem...)

Can somebody confirm this? Does somebody have an idea what to do?

Mike.

PMM
 

I think sdrconsole to low. The in no your rasp. Closer to reality.


On Sun, 22 Apr 2018, 22:44 matrixxx, <dm5xx@...> wrote:
Hi airspy,

Some strange behaviour with a higher noise floor while the antenna is not connected:
- if i connect the airspy hf+ directly to my pc, i see about -135dbm noise floor (using sdrconsole)
- if i connect the airspy hf+ to my spyserver running on a raspberry pi 3 (and 3b+, tested both), the noise floor is about 20dbm higher at -115dbm, which is A LOT!

To eleminate the interference i:
- connect to the spyserver via Wifi network (so no ethernet cable)
- attached a battery to supply the the raspberry and the airspy hf+ both with 5v enough power
- put the raspberry completely into a metal case

Nothing helped... So it seems that the noise is coming out of the USB socket (and/or the usb cable is bad, with just 2 cables connected, one of them has to be the problem...)

Can somebody confirm this? Does somebody have an idea what to do?

Mike.

prog
 

On Sun, Apr 22, 2018 at 02:44 pm, matrixxx wrote:
Hi airspy,

Some strange behaviour with a higher noise floor while the antenna is not connected:
- if i connect the airspy hf+ directly to my pc, i see about -135dbm noise floor (using sdrconsole)
- if i connect the airspy hf+ to my spyserver running on a raspberry pi 3 (and 3b+, tested both), the noise floor is about 20dbm higher at -115dbm, which is A LOT!
Welcome to the SDR world. The first thing you will learn (probably the hard way) is that SDRs are not Spectrum Analyzers. The scale has absolutely no meaning if you are comparing two different systems with different calibrations. That is, the digital gain difference has nothing to do with the RF noise whatsoever.

Also, for reporting problems: USE SDRSHARP and include screenshots with clear labeling.

To eleminate the interference i:

- connect to the spyserver via Wifi network (so no ethernet cable)
- attached a battery to supply the the raspberry and the airspy hf+ both with 5v enough power
- put the raspberry completely into a metal case
I'm not even sure you should interpret what you see as interference. Again => USE SDRSHARP.

Nothing helped... So it seems that the noise is coming out of the USB socket (and/or the usb cable is bad, with just 2 cables connected, one of them has to be the problem...)
The only valid question at this point is how does a reference signal show in both cases? What is its SNR? The bargraph in SDR# is actually an SNR-meter (and not an S-Meter). Use that.



Can somebody confirm this? Does somebody have an idea what to do?

Mike.

matrixxx <dm5xx@...>
 

hi,

thanks for reply. but I think I ran into the same problem like in that post:

Re: USB cable quality and QRM in a spyserver setup with an Airspy Hf+ and Raspberry PI3. 6 microUSB cable test.

#airspyhfplus 

 


The problem is the raspberry pi 3 USB port. the ports seem to emit this interference. unfortunately there seems nothing you can do about it. ferrite cores or double shielded cable seems not to work.
I know the difference between dbfs and dbm, and I took sdrconsole because u really can see the 20dbm there - and hear for sure.
-135dbm while direct connected to the PC, -115 using spyserver. both with no antenna. and u can hear the difference - with spyserver it's noisier.. so u have to decide for urself.. is the dx covered under the noise floor or not. to use the raspberry or not.

further ideas to eliminate the USB noise are welcome :)

Mike

 

prog
 

On Sun, Apr 22, 2018 at 03:35 pm, matrixxx wrote:
both with no antenna. and u can hear the difference
I know learning is hard as it starts by throwing away all your pre-existing convictions. But please save us some time: Use SDR# and do some screenshots.

jdow
 

Are you certain it is USB noise and not on board switching power supplies? Or are you simply blaming something popular to blame? Are you talking WGN level or spurious level? Being specific is a good thing.

{^_^}

On 20180422 15:35, matrixxx wrote:
hi,
thanks for reply. but I think I ran into the same problem like in that post:
Re: USB cable quality and QRM in a spyserver setup with an Airspy Hf+
and Raspberry PI3. 6 microUSB cable test.
#airspyhfplus
The problem is the raspberry pi 3 USB port. the ports seem to emit this
interference. unfortunately there seems nothing you can do about it.
ferrite cores or double shielded cable seems not to work.
I know the difference between dbfs and dbm, and I took sdrconsole
because u really can see the 20dbm there - and hear for sure.
-135dbm while direct connected to the PC, -115 using spyserver. both
with no antenna. and u can hear the difference - with spyserver it's
noisier.. so u have to decide for urself.. is the dx covered under the
noise floor or not. to use the raspberry or not.
further ideas to eliminate the USB noise are welcome :)
Mike

jdow
 

Terminate the inputs.
{^_^}

On 20180422 15:44, prog wrote:
On Sun, Apr 22, 2018 at 03:35 pm, matrixxx wrote:
both with no antenna. and u can hear the difference
I know learning is hard as it starts by throwing away all your pre-existing convictions. But please save us some time: Use SDR# and do some screenshots.

matrixxx <dm5xx@...>
 

hi Jdow,

thanks for your answer. no, I didn't want to blame something which is already known. I was just not good enough in searching the groups, I apologise for that.

the thread with the USB cable quality mentioned above is exactly the problem. unfortunately the guys there found also no real solution for the problem.

to eleminaze switching noise, I'm using a battery (and tried some other supplies earlier). it's also no radiation directly out if the raspi, that's why I put him into a metal case. to eleminate possible radion by Ethernet cable, I'm using WiFi. i tried also different USB cables. i did that unless only 2 cables are left: the power cord of the battery to raspi and the USB cable from raspi to airspy. it's also not a specific qrm. its easy like that:

- hf+ directly into the PC: -135 noisefloor
- hf+ into the raspberry: -115 noisefloor

I see also no problem not using sdr# ( like shd like it more, because I'm a c# software engineer :p).
Im using prg XYZ in both cases. If some uses sdrconsole, I'm sure he will agree a -20dbm signal is covering a -80dbm signal - if the measured values are really true ("calibration") or not. I'm sure, they aren't! but I also don't care if it's +-6dbm. we are talking about 20db difference using raspberry instead of a PC.
if sdrconsole is showing "nonsense" values at all, there is no reason to show diffent noise floors in an almost comparable setup: a computer is connected to the airspay... but if there are different noise floors, the only implicit rational reason is: one computer is behaving different to the other. so it could be an software issue or it cldn be the USB cable/port of the raspi. and that's what the guys of the "USB cable quality and qrm"-thread found: USB port noise from the raspi. use the best shielded cables u can get (I'm using original airspy USB cable) if you are lucky, it will be better, but be aware, even cores and ferrite clips might not work.

So the next steps will be: using a double shielded USB cable with 2 big ferite cores for blocking and pray that helps :p if not, I can decide if -115dbm noise floor is still ok.

Mike

prog
 

On Sun, Apr 22, 2018 at 11:09 pm, matrixxx wrote:
I see also no problem not using sdr#
At some point, you will make the link with the other "dBFS or not dBFS" thread that we started for you. So, yes, you are good for learning the hard way.

and that's what the guys of the "USB cable quality and qrm"-thread found: USB port noise from the raspi.
You are not even using an antenna. Which means you have a completely different setup => I conclude you have a completely different problem. Maybe you will face the same RPi QRM problem after understanding what's going on in your setup, but as I see it, this will take a lot of time.

As I always say in these situations: Good Luck!

Christof Proft
 

Hi,

try the Wuerth USB-EMC-Stick:

https://www.digikey.com/products/en?keywords=829993STICK

I don't have a RasPi 3 yet, I will buy an Odroid XU4. I have the Wuerth stick, the reduction of noise on a Windows PC' USB port is marginal. I think they are better schielded and blocked.

HTH

vy73
Chris

matrixxx <dm5xx@...>
 

Its always good to understand things :P As there is no antenna - if there is noise coming out of the Raspberry, its not radiated... I am with you dbFs is completely something different. And SDR# is cool. I also do not compare dbFS with dbm, bacause this is useless... Like comparing Apple with Peaches...

The only thing i like to understand is: Why the noisefloor values shown at sdrConsole connected to the PC are much much less than they are shown at the raspberry... Not even the exact hight is important,its just about why is it that much... dbFS is always dbFS - on the raspberry or on the pc... Around -95 dbFS...

If its not the USB-Port mentioned in the other post, maybe its the implementation of the linux drivers/spyserver-connection sdrconsole is doing in a different way than sdrconsole is doing. And you are completly right... The only thing to see if there is really an "interference" problem, is to use a low signal and measure the snr.I think i will do this and publish the results...

Some friends allready started to look into their setups to see if they have the same results like me. In a few more days we all know more.

M.

prog
 

On Mon, Apr 23, 2018 at 03:43 am, matrixxx wrote:
In a few more days we all know more.
Good luck. Ping me when you are at the notion of scaling.

marjohan@...
 

Hi!

One note based on my personal experience.

When you say you use "battery" as a power source for the RaspberryPi. Is this really a simple, non-BuckBoost-converter battery?

I also had some issues with the noice level even when using a battery to power my system. After I changed the power source to these "old fashioned" made with car battery and 7805 regulator I was able to get rid of some odd noice.

So called power banks they all use BuckBoost converters to get the normally used 3,7 volt cell lithium battery to provide that 5V output and that B&B converter is a loud piece of interference generator :-)

-Mac

matrixxx <dm5xx@...>
 

Hello Mac,

yes, its just a battery (with a filtered 7805) - it shows the same noise floor behaviour like all tested power supplies (including a real transformer supply, not a switching supply). So there is no difference between them. We are 4 People now investigating the same issue. First test with 10m beacon and an antenna show:there is a difference in SNR between Airspyhf+ connected to a pi and a airspyhf+ directly connected to the pc. The SNR on the Pi is at least 6db less when he is connected to the pc - at least there is a noticeable difference. Both sdrsharp and sdrconsole show that. Since sdr# is dbFS u just can see the the difference in SNR. And all see the same increase in the noise floor - and the increase is effecting the SNR. So it seems to be real. We are looking for someone with an Signal Generator to measure more exactly and we will take a look into the raspberys usb-connection...

Mike.

Martin Smith
 

The RPi has buck converts in them to generate all the required voltages (5V, 3.3V, 1.8V, and 1.2V), so even if one is running from battery the RPi will always be generating RFI. The RPi 3+ uses a new cheaper single custom made chip for this, the MXL7704 which provides four synchronous step down buck converters and a LDO.

matrixxx <dm5xx@...>
 
Edited

hi Martin,

thanks for info. that's interesting. this seems to be the reason why there was no difference between battery and all the other different power supplies we tried so far.
all measurements we ve be been doing since 2 days show that the issue seems to be the raspberry - probably the power system.
it's not dependent on the respberry 3 or 3+, it's happening on every raspberry 3. the noise floor is higher than using a PC. Hope someone will find the reason and a solution how to block the noise away.

m.

prog
 

On Mon, Apr 23, 2018 at 10:50 am, matrixxx wrote:
it's not dependent on the respberry 3 or 3+, it's happening on every raspberry 3. the noise floor is higher than using a PC. Hope someone will find the reason and a solution how to block the noise away.
Please keep the noise of this mailing list low as well. Your approach of not performing the designer's recommended tests is not only nonconstructive, but it also misleads many other users.