So, this is not really Arduino related, but general electronics.
Last week, my cable company send a technician because apparently our connection introduced massive noise into the network. This was after they switched frequencies, so presumably the noise was always there but on frequencies that where not used.
So, long story short, we could track it down to two power supplies (both brand models) in the general area (one next to the modem, one about a two meters away). Disconnecting those decreased the noise level (or rather the signal on his measurement device, not sure what it was) from almost 60dB to 30dB, which was apparently fine.
We essentially did a blind test where I unplugged the devices and he was sitting in the cellar giving me feedback. I only hat a glimpse on the display later but I could see that one caused a plateau on the left side, the other one 3 quite sharp spikes. I suspect the x axis was frequency, but I am not sure.
There were quite a few more power supplies, but the fact the two brand items caused the issue let's me suspect that those devices are actually within CE specifications and the shielding of the modem is not suitable for the new frequency range.
Does anyone have any idea or suggestion if it is possible to make some more investigations with a Rigol DS1054Z and the usual components one has lying around? I know the FFT of the Rigol is quite poor, but is there any chance to get something on the screen?
One of the power supply manufacturers already agreed to send me a new unit (Leicke, if anyone is interested), but they do not want the old one back, so they don't seem to be interested in investigating.
I have had some experience testing home made telephony devices which I have connected to my cable modem/router and noticed that some older mobile phones could interfere with it if they were too near making the signal unusable. The telephony devices had a 600 Ohm isolating transformer and I guess this was picking up unwanted signal radiation.
If you are testing with an oscilloscope, maybe try with a copper wire coil and see what it picks up.
6v6gt:
If you are testing with an oscilloscope, maybe try with a copper wire coil and see what it picks up.
Thought about something like that.
Rough theory is clear, it's rather about the details: What frequency range would I look for, or what coil design would give me the broadest band width?
Does anyone have any idea or suggestion if it is possible to make some more investigations with a Rigol DS1054Z and the usual components one has lying around?
Hmm tricky, the usual culprits are 100MHz+, 150Mhz seems to be a favorite frequency some switchmode power supplies put out (it's not the switching frequency of the PSU, it's the ringing due to switching).
Lower frequencies tend not to couple as easily anyway.
So, with a 50MHz 'scope alone you are not going to see much and without some external circuit to pick up RF frequencies, I suspect what you will get will be swamped by mains 'hum'.
You may be able to hear some sidebands with a manually-tuneable FM radio - look between stations for buzzing noises.
If you can detect something, by whatever method, you can then try some clip-on ferrite suppressors next to each PSU where the mains goes in and where power comes out.
I highly recommend Henry Ott's book EMC engineering, its all in there, IME no matter what your issues (I have had a lot) its in there
The OP doesnt give many details, what frequencies are we talking about?, I assume its conducted as the noise seems to be measured external
Many people dont realise that the vast majority of EMC emissions is common mode and thats why capacitors and standard filters dont do a thing to reduce the emissions. You almost always need to add common mode impedance. Because its common mode then a scope with an FFT isnt going to help. Using a spectrum analyser gives a clue but background noise unknown antenna losses variable distance from probe to unit means you cant do anything but baseline i.e take a measurement at fixed conditions and see how far the fix reduces things
Conducted emissions are the easiest to reduce, it depends on the frequencies, 30dBuV is a hell of a lot for one PSU many limits are only 50dBuV
I once owned a Zyxel wireless router (reasonably name brand) I had some issues with my Wifi connections but at the time I thought it was normal. I received a notice one day that they had a bad batch of power supplies and they were sending me a new one. The bad ones were emitting excessive noise.
Received the new supply (aka wallwart) and my Wifi was all of a sudden much better.
Moral: Even good companies can have some bad supplies.
An interesting test would be to contact your name brand supplier and tell them their product was emitting noise and see what they say.
The OP doesnt give many details, what frequencies are we talking about?, I assume its conducted as the noise seems to be measured external
As I said, I don't know. I couldn't see much of the display. Maybe someone who knows what kind of test equipment a cable technician uses can say anything about the units on the display. I guess it was not in the easy range, since they apparently just added frequencies and that seemed to have made the issue apparent.
JohnRob:
An interesting test would be to contact your name brand supplier and tell them their product was emitting noise and see what they say.
Already did, it's in the OP. Good: They are replacing it. Good/Bad: They don't want the unit back, so they are not investigating.
A Realtek SDR (RTL2832U) has a frequency coverage of 64 - 1700 MHz and can be used as a poor mans spectrum analyzer.
About £5.
Anything special I have to take care of? This fine?
I do own the same RIGOL, the 1054Z and it can very easy be hacked to work perfect to 100Mhz and you get all the special software unlocked as well (like for instance the I2C decoder) RIGOL knows about this hack since it has been placed online since 2014 and doesn't mind it, since the sales of the 1054Z skyrocketed since it was publisched. This being said, as an intro.
It is very hard to measure those frequencies with a scope limited to only 100Mhz, they tend to be situated in a higher frequency range, plus the hardware required, the spools, the number of windings, and sometimes they use antennas with very specific lengths are very expensive, I am talking about thousands of $. If you want to learn more have a look at the EEVBlog the forum and on Youtube, Dave made a video where he measured those interfering noises, using the special equipment. I have searched, but can't find the specific video I mean.
I hoped to have been helpful for a bit, even if it is only to let you know that you can safely hack your 1054Z to 100MHz and get all the special software working. Like the different triggers.
Already did, it's in the OP. Good: They are replacing it. Good/Bad: They don't want the unit back, so they are not investigating.
Sorry I missed that in my initial read.
As suggested, EMI/EMC is not for the faint of heart. It is somewhat complex and requires special equipment and a shielded room to test in.
As much of the switching noise tends to be broadband, For my home projects I simply use a cheap AM radio. Bring it near the circuit and if you can detect the interference (usually at the lower AM frequencies) you might have a problem. If you can't detect any interference you are likely to be OK. Where OK means you likely won't cause any disruption of other devices.
sometimes they use antennas with very specific lengths are very expensive, I am talking about thousands of $
Well that shouldn't be an issue, since the bare wall socket worked very well to pick up the noise (It was not the cable, since the technician could screw something on the F connector to shield is and the noise was gone)
Also, of course my Rigol is hacked. I didn't mention because I thought that is the standard state of a hobbist's Rigol scope g
That will work, but those with the R820T tuner are generally considered to be a bit better. See here: http://rtlsdr.org/.
The antennas that come with these are univerally not particularly good for broad band work, but to get a qualitative scan of high powered interference, it shouldn't matter.
(It was not the cable, since the technician could screw something on the F connector to shield is and the noise was gone)