PoE noise

[Matthew]

From reading this group there seems to be a fairly diverse range of networking hardware in use. If perhaps you think you are suffering noise problems the following may be of interest.

I have setup a Ubiquiti modem to get to Ethernet out to my garden. At the remote end I have a Ubiquiti PoE adapter, Ethernet cable to HL2, Ethernet cable to Ubiquiti modem.

I have had problems at power > 50 W causing the HL2 connection to drop out over this link. Connecting a dummy load to the HL2 at the remote end had no drop outs, and connecting from a laptop straight the the HL2 caused no drop outs.

I put a problem solver on 3 loops of the Ethernet cable to the HL2 and I can now TX with no connection loss. There was also about a 6 dB drop in noise (depending on the band).

I wanted to try 10m this weekend for the ARRL-10 contest. I was struggling to hear very much but RBN spots suggested I should be able to hear a little more than I could. I removed the problem solver ferrite from the Ethernet cable and put it on the mains lead to the Ubiquiti PoE adapter with the cable looped through the ferrite 5 times. I saw a 13 dB drop in noise.

I still had to fit a ferrite on the Ethernet cable without a clamp. I went with 3 turns through this one.

I'm not convinced that good quality shielded Ethernet cables would have solved this.

73 Matthew M5EVT.

[Mooneer Salem]

Hi Matthew,

I too have had serious RFI issues with my Ubiquiti UAC-AP-HD, especially on 40 meters. I ultimately had to do the following to get the noise from it below my local noise floor:

1. Replaced the runs from the AP to the PoE injector and from the injector to the switch with CAT 7 shielded Ethernet cable.

2. Added four clip-on ferrites on the Ethernet cables (one at the AP, two at the injector and one at the switch) with sufficient diameter to wrap the cable with as many turns as possible. 3' CAT 6 jumpers using thin cabling helped significantly here.

3. Swapped out the injector it came with for one that uses an external power supply (this one is what I'm currently using). I also wrapped the external power supply's AC lead around a toroid core and the DC side around another clip-on ferrite.

However, I suspect most people won't need to go to that extent as my antenna/equipment positioning might be particularly bad; I live in a condo building where the MFJ magloop that I have mounted on the roof is about 5-6 feet above the ceiling mounted AP and another 3-4 feet above the switch and PoE injector. There might also still be noise from it but I'm unable to particularly tell given my noise floor is now around S4-5ish on 40 meters and S2-3 on 20 per my backup IC-7200. Hope this helps others!

Thanks,

-Mooneer K6AQ

--
You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/e7a66b3c-a11f-4b07-af1e-1c0e98925b9fn%40googlegroups.com.

[Steve Haynal]

Hi All,

Thanks for the helpful suggestions. I've seen noise due to ethernet too. I think some "bad" devices introduce RF noise onto the ethernet cable and if the cable if relatively long it becomes an antenna. Then if your HL2 antenna is in close proximity, you will pickup up the noise. This is also true for many wall wart power supplies. I've found that they inject RF to my house AC power wiring which can be picked up by my attic antenna. You can tell if the noise is coming in to the HL2 through the antenna by removing the antenna connection. Usually the noise goes away with a disconnected antenna. I have seen very few instances of RF noise entering the HL2 through the ethernet jack or power connection. Much to my family's annoyance, I will go around the house removing power supplies or connected devices and flipping circuit breakers to hunt down noise. Unfortunately, often the noise comes from my neighbors. 

73,

Steve

kf7o

[radi...@mail.com]

Agree totally with you Steve. I've had a lot of noise issues to contend with in my own house and every one of them so far has turned out to be radiating switched mode PSUs, mostly of the "wall wart" plug-in variety so common with most modern kit. 

There is also not necessarily immunity when using "named" brands either. One of my worst radiators was the wall wart PSU for a Netgear switcher, with Netgear branding on the PSU. Also agree not really seen any issue caused by network cabling. I have CAT5e cabling (installed few years ago) all over the house and no noise detected from that. Certainly IMHO first stop is to check all the PSUs on the system and change them out for a different brand if necessary. Just for info, I don't find all SM PSUs noisy. I have some undetectably quiet ones too.

Steve, you are not the only one to annoy family by turning off kit (or the main power to the whole house) when trying to track a noise source, though if possible I try to wait until they are all in bed!

73

Max

[murphy edward]

Hello Steve,

Have you ever seen noise coming OUT of your HL2 ? 

Please check out the wide band radiation from the HL2 power supply leads. The on board switching regulators are producing RF energy that is detectable over 500 metres from the unit ! 

Yes, it is likely that it is not just my HL2 example that is capable of widespread interference to other sevices. But possibly all HL2 units built to date ? Unless later units ( than version 5 ) have addressed this subject ? 

In the initial design conception phase of this project  during your "Project Definition Study",  was the subject of CE approval addressed ? 

Was the appropriate design elements added to give the necessary level of attenuation of "on board" generated noise being conducted/ radiated to the either ?

It may be prudent to conduct a wideband measurement of the radiated energy from the HL2 and implement corrective action where necessary.  Has this ever been done during the very early stages of the design process ? 

I had to add modifications to my HL2 to reduce the radiation to acceptable limits.  

Regards

Ed GM3SBC  

--

You received this message because you are subscribed to the Google Groups "Hermes-Lite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hermes-lite...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/1876cd0d-d391-413a-aade-4c94543f14a0n%40googlegroups.com.

[Mooneer Salem]

This is hardly a scientific test but I just tapped the whip that came with my tinySA on my HL2's power lead and didn't see anything extremely crazy. The attached image shows the tinySA's display (set to show 1.8-30 MHz) while the whip was touching the lead. My HL2 is from the batch that just sold out, FWIW.

On that note, I've been playing with some interesting "industrial/medical" isolated power supplies from TRACO as part of my Wi-Fi magloop tuner project. They're pretty expensive at $20-30 each in low quantities on Mouser but the TVN-3 in particular along with appropriate pi filtering on the input and output don't seem to radiate anything of importance on the ham bands that my MFJ loop is capable of using (and what is being radiated I imagine is easily suppressed with a clip-on ferrite). There might be additional improvement to be had by hooking the IC's metal case to one of the grounds--since the tinySA does pick up stuff if the whip is within a few cm of it--but I haven't had the need to do so yet.

-Mooneer K6AQ

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/566192968.2260592.1607993212350%40mail2.virginmedia.com.

[Steve Haynal]

Hi Ed,

No, I have never seen noise coming out of the HL2 power connection, and I have measured. Maybe it is your power supply? Is your board a custom build? Maybe there was a bad part substitution. One of the benefits of an open hardware project is that we had many early adopters and testers. Users would have reported if noise generated was significant enough to matter. Users do see the switching frequencies on 160M due to magnetic coupling with the N2ADR board. The HL2 solves this by having the novel ability to control the switching frequency and move the spurs. They are rarely above the noise floor on 160M anyways.

The HL2 is an open hardware/open source project worked on part time by a few people. It is a kit sold at a very low cost and low volume for experimenters. There are no plans or expectations for the rigors such as CE approval that you might see with a commercial product. Even so, the HL2 is pretty well tested and a good decent SDR for a very low price. This is similar to the softrock, RS-HFIQ and ucx/qcx+ which also do not have CE approval.

For full disclosure to the list, you are a maker and vendor of a competing commercial and proprietary SDR. Without specifics about the the noise you measured (which frequencies? how severe?) and no details about your modifications to reduce the noise (what exactly did you do?), I am not sure if you are trying to help the HL2 or spread FUD (fear, uncertainty and doubt). Please clarify your intent.

73,

Steve

kf7o

[Steve Haynal]

Hi Ed,

Thanks for the additional information. This does make sense. This type of technical feedback is what has improved the HL2 over the years. I think most noise checks have been focused on the RX frequency range of the HL2, to ensure it is not interfering with itself. I will have to set up my SA and take a look at the 400-500MHz and higher range on the power supply. Perhaps just an easy to install capacitor across Vsup to GND at one of several available spots on the PCB would go a long ways to reducing noise in this frequency range.

73,

Steve
kf7o

  

On Tue, Dec 15, 2020 at 3:31 AM Ed Murphy <ed.g...@blueyonder.co.uk> wrote:

Hello Steve,

The radiation from the power lead was observed within the 400 to 500 MHz range for my particular test. The circulating currents at the input to the switching regulators were identified as the source.

The capacitors at the input to the switching regulators, ( C17,18,23,24,33,34,and C86) are all high value capacitors that are likely to have low SRF's and may not operate well as capacitors within the higher frequency spectrum.

I found that there was radiation on the 'PCB Input Track' from the CN2 through Q1 Source. This noise energy was then feed out to the + Ve power supply lead. As it happened, the length of the power supply lead provided  a relatively good 'match' in the 400 to 500 MHz band, hence the external radiation levels experienced.

To satisfy my particular requirement I cut the PCB track feeding the CN2 Power Supply connector as close as possible to the CN2 centre pin connection. I fitted a ferrite inductor across the cut track. On the rear of the PCB I fitted a high SRF decoupling capacitor directly across the earth and live pins of the CN2 connector.

It was important to have this LC filter placed as close as possible to the power supply output node point otherwise there is the possibility of radiated / conducted noise being re introduced within the enclosure.

The board I have is a 'build 6'.

All my power supplies are Lambda linear supplies. - very quiet.

This modification worked for my requirements.

I did not do an entire sweep of the noise spectrum out to higher frequencies at the time as other matters were pressing.

I hope the above makes some sense.

Regards

Ed GM3SBC

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/92949264-6ca6-4c49-bdf4-7265adacfc4bn%40googlegroups.com.

[ron.ni...@gmail.com]

Hi Steve,

As a precaution, I had a ferrite bead on the power supply lead from my old Radio Shack linear regulated 12V supply to an HL2.

I removed the bead and wrapped the power supply lead around a small antenna connected to an RTL-SDRv3, then swept the frequency range from 400 to 500 MHz with the RTL-SDR.  I see two very small spikes in the spectrum around 500 MHz (an uncalibrated frequency), one about 10 kHz lower, another about 30 kHz higher, that are not present until after power is applied to my HL2,  The spikes bounce up and down a few times initially, and become solid just after the HL2v9 acquires its DHCP address.  They seem narrow band (as far as I can tell with the RTL-SDR).  I don't have a good reading of the levels, but they are clearly above the (high) noise floor in my office environment.

I'll put the ferrite bead back on my power supply lead until we can figure out if/where a bypass cap might be needed.

73,

Ron

n6ywu

[Steve Haynal]

Hi Ron,

I haven't looked with my spectrum analyzer yet. I am hoping that some of those using the HL2 with a transverter might report back with what noise they see and how serious it is. Ed's solution sounds like the right way to go and is not very expensive or difficult.

73,

Steve

kf7o

[Steve Haynal]

Hi Group,

I have opened a github issue to track this:

https://github.com/softerhardware/Hermes-Lite2/issues/153

73,

Steve

kf7o

[Hamish Kellock OH2GAQ]

Hi Steve,

I don't have any data regarding the noise in the 400 to 500 MHz region but have some considerable operational experience with the HL2 as a 28 MHz tuneable IF for 144 MHz transverters, which are used in turn as tuneable IF's for higher frequency microwave transverters (for example at 2.4 GHz and 10 GHz). I found a Quisk screen dump from some time ago (Build 7 or 8 I think), which shows a fair part of the 2M band (covering a couple of beacons in Helsinki area). I didn't find any significant spurs over the 144.28 to 144.64 shown in this plot, and also from 144.0 to 144.28 MHz. And haven't found anything in the 10 GHz Rx band for Amsat Oscar 100 that I can attribute to the HL2+R. Pi 4B+rest of the Rx. and Tx. chain. If I have a chance to measure the power feed with a SA. I will do so, but presently are nowhere near the shack. All the 144M plus stuff is in pretty well shielded enclosures, anyway.

73,

Hamish, OH2GAQ

[Steve Haynal]

Hi Hamish,

Thanks for the feedback. I have not yet tried to make my own measurements, but the solution is simple enough for any future HL2 revisions.

73,

Steve

kf7o

[Steve Haynal]

---------- Forwarded message ---------
From: F5NPV Amateur Radio <didier....@gmail.com>
Date: Sat, Jan 2, 2021 at 2:45 AM
Subject: Re: PoE noise
To: Steve Haynal <softerh...@gmail.com>

Hello,

According any possible POE or Ethernet Noise , i use to wrap my ethernet cable and power line with 6 to 8 turn around a FT240-31 toroid. In addition the overall shack is grounded with a dedicated ground line and pole and all my coaxial feeder are equipped with a dedicated CMC balun. My place is located close to a very high voltage power line and without all of above , my noise level is very high. Basically all my wire or cables are equipped with 31 ferrites or toroids and feeders with CMC systems. it cost a bit of money but provide me some great results.My noise floor is very low currently.

According any POE noise i did some quick check with an analyzer and did not notice any spurious noise or spike due to lan connectivity. Maybe in some cases some IMD from other systems may create some spurious or spikes and intermodulation products which are always very tedious to investigate.

73s Didier

To view this discussion on the web visit https://groups.google.com/d/msgid/hermes-lite/245ce509-6f0d-4cab-a9dd-783b6af9805fn%40googlegroups.com.