Author Topic: DIY EMC test cage  (Read 5306 times)

0 Members and 1 Guest are viewing this topic.

Online nctnicoTopic starter

  • Super Contributor
  • ***
  • Posts: 26752
  • Country: nl
    • NCT Developments
DIY EMC test cage
« on: September 01, 2022, 08:47:11 pm »
As part of hunting a nasty EMC problem I embarked on trying to create a shielded environment that would allow me to make some kind of measurements on a device that fails the radiated emissions testing. There is just too much interference around to make a measurement without any form of shielding. A problem is that the emissions come mostly from the wires and these can't be shielded or filtered so I need quite a big setup to include the wiring as part of the measurement.

I did find conductive cloth on Aliexpress but making a huge tent with it would be costly an no guarantees on whether it would work. I thought about different solutions and came up with the idea of using wire mesh that is used for fences or cages. At least this would be self-supporting and simplify the design because it shouldn't need a frame. So I went to the hardware store and bought some bird cage wire mesh and started folding it into a cage:



This is the nearly finished 'product'; some initial testing looked promising enough to go forward. After I added the wooden edge frame I was quite content with how sturdy the cage felt. But that feeling didn't last long. I put some weight on top to keep the mesh for the lid flat and then it decided to collapse in on itself at the left side. Curses!  :rant: I managed to straighten it out but it will need some reinforcement later on.

The size is 60x80 by 90 cm high. This is a size I have a storage space for.
I used aluminium tape (real aluminium so it is conductive) to get rid of the sharp points of the wire mesh. I laced the bottom and the side together with steel wire so there is a lot of contact between the mesh.

The shielded chamber 101 says that the shield of signals coming in needs to be fully connected to the wall of the chamber so I made this with some bulkhead connectors and a piece of copper plate:



For other signals a copper tube to feed them in is the recommended option so I added a piece of 22mm copper pipe. I choose 15cm as a reasonable length. Last but not least I soldered the copper plate into the mesh.

Ofcourse I also needed some kind of antenna. I opted to make a simple dipole and found a website saying that making a dipole wide, makes it broadband. Just slap something together and see what it does. I found some phenol copper clad in my stash with PCB material and made an antenna which is nearly 90cm high. The plates are 10cm wide.

Let's hook things up:


Meanwhile I made a lid as well that fits nicely around the cage. I used a piece of PVC pipe to hang cables from. The mesh allows to tie cables down as well.



The cage provides a shielding of around 25dB at relatively low frequencies:

The yellow trace is with the lid open, the blue trace is with the lid closed. The blue trace is at the noise floor of the spectrum analyser in most places. The FM radio stations remain clearly visible though but for the measurements I wanted to make most urgently, these are not in the way.

Working close to the noise floor of the spectrum analyser did cause some challenges. Even with the preamplifier on, I just missed a bit of dynamic range. So I bought a simple amplifier from Amazon.

I used a USB cable and a ferrite core (with only the power wires wrapped around it) to power it.

The amplifier itself is build around a SPF5189 which turns out to be a very low noise amplifier.


Now this amplifier has a noise floor itself but it seems to give an extra 5dB to 10dB of dynamic range. Enough to pull the signals a bit more out of the noise.

Same signal measured with (top trace) and without (bottom trace) amplifier:


All in all I'm not dissapointed. Unfortunately the cage isn't big enough to do real far field measurements from a few meters away. And the antenna leaves much to be desired for sure but this cage did allow me to try a few modifications and gather some data to justify going to a real EMC lab or needing to do more work on the circuit.

A future enhancement could be adding a finer mesh or conductive cloth to improve shielding. I have found a report that indicates a finer mesh leads to better shielding at lower frequencies. Or maybe I opt to make a large tent from conductive cloth which I can put up in the living room temporarily  ;D
« Last Edit: September 01, 2022, 09:13:50 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline wilhe_jo

  • Regular Contributor
  • *
  • Posts: 174
  • Country: at
Re: DIY EMC test cage
« Reply #1 on: September 02, 2022, 10:35:49 am »
Ever thought of building a crawford cell?

That's basically a big coax... similar to GTEM/TEM cells, but easier to build than GTEM cells and compared to normal TEM cells they benefit from the shielding.
For the lower-end of the spectrum, they should be fine for you.

A mesh is always a bit of a compromise. Sheet metal is the proper thing...

73
 

Online nctnicoTopic starter

  • Super Contributor
  • ***
  • Posts: 26752
  • Country: nl
    • NCT Developments
Re: DIY EMC test cage
« Reply #2 on: September 02, 2022, 04:29:32 pm »
I have been pondering on a Crawford / TEM cell. The problem is that I don't know how large it would need to be to include the device + a decent length of wires AND whether it is even worthwhile to embark on building it. Ideally I would like to do measurements that represent a far field situation on a setup that is 1 meter long. Looking at the specs for Tekbox open TEM cells, I estimate I would need to build one that is 2.5 to 3 meters long. Actually an open TEM cell with external shielding looks more practical to me because it is easier to access the DUT (on top of being simpler to build).
« Last Edit: September 02, 2022, 04:36:17 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline coppercone2

  • Super Contributor
  • ***
  • Posts: 9234
  • Country: us
  • $
Re: DIY EMC test cage
« Reply #3 on: September 02, 2022, 05:13:51 pm »
I kinda wanna build this, I recommend putting wood stain on the wood structure before assembly though to make it more durable and visually pleasing, and copper instead of aluminum tape, and a vanity garden fence under the wire mesh to make it better structurally, or perhaps over it, so that its harder to deform the mesh.
« Last Edit: September 02, 2022, 05:15:55 pm by coppercone2 »
 

Offline jonpaul

  • Super Contributor
  • ***
  • Posts: 3318
  • Country: fr
Re: DIY EMC test cage
« Reply #4 on: September 02, 2022, 05:46:19 pm »
Use copper screen, not galv mesh steel.

Test with local noise RF sourcs on, off eg any electronic lLAMPS LED, CFL, HID

j
Jean-Paul  the Internet Dinosaur
 
The following users thanked this post: pardo-bsso

Offline coppercone2

  • Super Contributor
  • ***
  • Posts: 9234
  • Country: us
  • $
Re: DIY EMC test cage
« Reply #5 on: September 02, 2022, 06:11:01 pm »
I would wrap copper over the galvinized steel because its too easy to puncture a thin screen like that and totally destroy it
 

Offline mag_therm

  • Frequent Contributor
  • **
  • Posts: 633
  • Country: us
Re: DIY EMC test cage
« Reply #6 on: September 02, 2022, 11:12:25 pm »
Navy had screened rooms, but for big stuff, a way to do approximate tests was to load the DUT and test gear along with a portable generator  on a truck and drive to a natural gully with hills all around. I suppose about 4am would be the best time for HF. Usually would be lucky to get there by mid morning after getting organized and finding all the gear and cables , passes to get stuff out the gate etc.
 

Offline coppercone2

  • Super Contributor
  • ***
  • Posts: 9234
  • Country: us
  • $
Re: DIY EMC test cage
« Reply #7 on: September 03, 2022, 04:24:24 am »
 

Offline coppercone2

  • Super Contributor
  • ***
  • Posts: 9234
  • Country: us
  • $
Re: DIY EMC test cage
« Reply #8 on: September 03, 2022, 04:26:52 am »
seriously trying to find some stupid cable and getting papers signed at 4am, what a nightmare

let me guess you could not pre-package it the week before because it was in use right?
 

Offline mag_therm

  • Frequent Contributor
  • **
  • Posts: 633
  • Country: us
Re: DIY EMC test cage
« Reply #9 on: September 03, 2022, 06:06:43 am »
I do recall a funny thing happened. Even though long time ago, better stay silent.

I mentioned this as a possibility of time and place  for nctNico, to drive out to low lying rural area and work at night.
 To maybe get a lower ambient, not knowing how quiet he needs.
 

Offline wilhe_jo

  • Regular Contributor
  • *
  • Posts: 174
  • Country: at
Re: DIY EMC test cage
« Reply #10 on: September 03, 2022, 07:16:07 am »
I have been pondering on a Crawford / TEM cell. The problem is that I don't know how large it would need to be to include the device + a decent length of wires AND whether it is even worthwhile to embark on building it. Ideally I would like to do measurements that represent a far field situation on a setup that is 1 meter long. Looking at the specs for Tekbox open TEM cells, I estimate I would need to build one that is 2.5 to 3 meters long. Actually an open TEM cell with external shielding looks more practical to me because it is easier to access the DUT (on top of being simpler to build).

Actually, a Crawford cell is a tem cell with shielding... However, your size requirement is kind of a problem.

I have a 5m long GTEM cell... Max eut size is somewhere between 20 and 25cm (dimension of a cube).... But that thing is fine up to some GHz...

On the other hand, the usual 3m distance in an anechoic chamber does not really give you far-field results in the lower frequency range either.

Btw, I did some enquiry on EMC chambers in China. If you have the space and your're willing to assemble it yourself, you get a basic chamber for 100-150k...

The next best thing would be to screen a room in a basement... I have one where ambient is right at the emission limits for industrial stuff. So the 20dB of your mesh would give me sufficient headroom even for residential testing.

However, a big room without ferrites gives you quite some troubles with reflections...

So all in all, radiated emission testing is nothing you can do without serious investments ( money, real estate and time)

For large stuff, I tend to put the antenna as close as 1m. That way I get higher emission levels, but converting limits, re-calibrating antennas,... took quite some time.
To "optimize" ambient noise, I turned my Yagi 360°, to find the worst direction, turned it 180° and moved the EUT accordingly. This helped a lot for my situation.
 
73
 

Online nctnicoTopic starter

  • Super Contributor
  • ***
  • Posts: 26752
  • Country: nl
    • NCT Developments
Re: DIY EMC test cage
« Reply #11 on: September 03, 2022, 12:09:36 pm »
I have been pondering on a Crawford / TEM cell. The problem is that I don't know how large it would need to be to include the device + a decent length of wires AND whether it is even worthwhile to embark on building it. Ideally I would like to do measurements that represent a far field situation on a setup that is 1 meter long. Looking at the specs for Tekbox open TEM cells, I estimate I would need to build one that is 2.5 to 3 meters long. Actually an open TEM cell with external shielding looks more practical to me because it is easier to access the DUT (on top of being simpler to build).

Actually, a Crawford cell is a tem cell with shielding... However, your size requirement is kind of a problem.

I have a 5m long GTEM cell... Max eut size is somewhere between 20 and 25cm (dimension of a cube).... But that thing is fine up to some GHz...

On the other hand, the usual 3m distance in an anechoic chamber does not really give you far-field results in the lower frequency range either.

Btw, I did some enquiry on EMC chambers in China. If you have the space and your're willing to assemble it yourself, you get a basic chamber for 100-150k...

The next best thing would be to screen a room in a basement... I have one where ambient is right at the emission limits for industrial stuff. So the 20dB of your mesh would give me sufficient headroom even for residential testing.
Now you are entering into a grey area  ;D

Let me throw some thoughts out on the table... If I need to spend over >1k euro in materials and a considerable amount of time, then it is more efficient for me to go to an EMC lab. On top of that, I'm quite space constrained as well. The primary purpose to have some testing facilities in my own lab is to be able to make comparative measurements to assess the effectiveness of changes to a circuit. The shielded cage I made allowed me to do that because it has enough attenuation of outside noise for the particular case I had to work on. I'm the first to admit it is far from flawless though.

A TEM cell looks like a good next step to take for a more permanent solution that also seems to be closer the kind of results I would get from testing at an EMC lab. With my mesh cage I have not been able to see the exact same peaks as during the radiated emissions pre-scan testing. What I did is modify the circuit so frequencies that are in the same band are suppressed and measurements in the mesh cage seem to confirm that but the confidence level of the measurements isn't high. If a TEM cell can get me higher confidence levels then it could be a worthwhile investment. Not only for fixing EMC problems but also as an extra service towards my customers. Typically I like to do EMC pre-scan tests (as good as I can) for devices that I develop to get some feeling whether it radiates a lot or not. Though a lot of devices I develop are never going through an official EMC test because the volume is too low. Another advantage of a TEM cell is that it can be used for immunity testing as well so that would be a plus. But at this point I'm not 100% sure of what a TEM cell would give me in terms absolutely measurement accuracy. If I can get within 6dB compared to a real EMC test, then I would be happy already. I see that Tekbox claims that using their TEM cell can give a good indication whether a device would pass or fail radiated emissions testing limits.

Meanwhile I have found some interesting documents on TEM cells:

A report co-written by Crawford which has some hands on formulas for building a TEM cell and some experimental results:
https://nvlpubs.nist.gov/nistpubs/Legacy/TN/nbstechnicalnote1013.pdf

And a report for a DIY TEM cell (however I question the correctness of the formulas used):
https://www.ewh.ieee.org/soc/emcs/acstrial/newsletters/summer08/pp2.pdf

I have plugged some numbers & formulas into an Excel sheet in order to play with them a bit. It looks like a 2 meter long TEM cell (almost sigar shaped) would work for me. Probably it will need some shielding at the sides and / or partly closed sides to have mechanical strength. Due to the length the resonance frequency will probably be quite low though.

I mentioned this as a possibility of time and place  for nctNico, to drive out to low lying rural area and work at night.
 To maybe get a lower ambient, not knowing how quiet he needs.
I'm in a country that has been shoved flat as a pancake by an ancient glacier so there aren't any valleys. Also due to dense population there are cell towers everywhere, wind turbines (which I assume have some power electronics on board) plastered allover the place and there is a huge AM transmitter station nearby.
« Last Edit: September 03, 2022, 08:46:49 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Online radiolistener

  • Super Contributor
  • ***
  • Posts: 3282
  • Country: ua
Re: DIY EMC test cage
« Reply #12 on: September 03, 2022, 10:08:13 pm »
What kind of field you're want to reject? Electric or magnetic? Or electromagnetic plain wave?

At low frequency the most complicated thing is to reject magnetic near field.

I think you're needs to use thick layer of solid iron instead of metal mesh...
And keep the distance between cage walls and test equipment as far as possible in order to reduce near field coupling...
« Last Edit: September 03, 2022, 10:10:34 pm by radiolistener »
 

Online nctnicoTopic starter

  • Super Contributor
  • ***
  • Posts: 26752
  • Country: nl
    • NCT Developments
Re: DIY EMC test cage
« Reply #13 on: September 05, 2022, 06:04:01 pm »
Magnetic fields are not the problem. These die out pretty quick anyway. My problem are nearby transmitters and electrical noise that is typically present in a city. I did a scan up to 1GHz and LTE cell towers come in loud and clear inside the mesh cage. Not a surprise though.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline max-bit

  • Frequent Contributor
  • **
  • Posts: 668
  • Country: pl
Re: DIY EMC test cage
« Reply #14 on: September 05, 2022, 08:39:33 pm »
A very average idea.
It is not known to what extent this will be suppressed by the given frequencies.
The cage itself may, for certain frequencies, show the phenomena of (internal) reflection resonances, so the parameters are very undefined.
Building such a chamber at home is difficult and, for sure, expensive.
It was better to do it from, for example, thin aluminum sheet.
But there is still the question of internal reflections (and resonances, standing waves and such) there are special materials to dampen such hovers, unfortunately they are very expensive.
 

Online nctnicoTopic starter

  • Super Contributor
  • ***
  • Posts: 26752
  • Country: nl
    • NCT Developments
Re: DIY EMC test cage
« Reply #15 on: September 09, 2022, 12:42:54 am »
Yuk... Still a no go at the EMC lab. I'm strongly suspecting there is a problem in the part of the test setup outside the chamber which ends up injecting noise into the chamber. If not then then I'm going to built a TEM cell that is large enough to hold the device under test and the cabling.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Online radiolistener

  • Super Contributor
  • ***
  • Posts: 3282
  • Country: ua
Re: DIY EMC test cage
« Reply #16 on: September 10, 2022, 01:54:26 am »
Magnetic fields are not the problem. These die out pretty quick anyway. My problem are nearby transmitters and electrical noise that is typically present in a city.

RFI from nearby transmitters in GHz and sub-GHz band can be easy shielded with a metal box, because they are usually in a far field region. But there are a lot of noise from SMPS, LED lamps and other home electronics in the near field region, and it's not easy to fight with them because they are below 10 MHz. And even worse, they have RF leakage into mains, so you're needs to use mains filters against it.

The same brand LED lamps can have RFI at different frequencies. But if they from the same part, their RFI peaks almost the same, but also have some frequency shift. So, if LED lamp produce RFI at unwanted frequency, you can try to replace it with another one. In such way you can put RFI outside spectrum band of interest. But if you want to clean noise floor, you're needs power off all LED lamps in your home  :)

You can catch some LED lamp which can produce very strong RFI noise. So, you're needs to test all your LED lamps.
« Last Edit: September 10, 2022, 02:08:11 am by radiolistener »
 

Offline tautech

  • Super Contributor
  • ***
  • Posts: 28136
  • Country: nz
  • Taupaki Technologies Ltd. Siglent Distributor NZ.
    • Taupaki Technologies Ltd.
Avid Rabid Hobbyist
Siglent Youtube channel: https://www.youtube.com/@SiglentVideo/videos
 

Online mawyatt

  • Super Contributor
  • ***
  • Posts: 3194
  • Country: us
Re: DIY EMC test cage
« Reply #18 on: September 15, 2022, 04:37:25 pm »
Back in 2000 we had to build a "special room" that had limited access physical and electromagnetically. The "room" hosted a lab and conference room, and the usual cost for these types "rooms" was daunting!!

We ended up using the insulating type drywall equivalent (8' by 4') which was a foam type with a thick aluminum outer covering for the walls, sub-ceiling and sub-floor. The 8' by 4' sheets were covered with another layer of thick aluminum foil and the seams were covered with a conductive aluminum tape, and everything was bonded with a conductive bonding epoxy and conductive screws. We saved a bundle which the executives took credit for and pocketed nice additional bonuses, typical of how engineers/scientist help enrich the company executives while suffering no real financial reward, other than a stinking plaque |O

Anyway, this material & technique might be useful for fabricating a DIY shielded enclosure/room on a budget.

Best,

Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 
The following users thanked this post: nctnico

Online rhb

  • Super Contributor
  • ***
  • Posts: 3476
  • Country: us
Re: DIY EMC test cage
« Reply #19 on: September 15, 2022, 05:04:40 pm »
I’d like to suggest you use galvanized HVAC duct metal.  It’s cheap and effective.  You’ll want to solder the seams or at least cover them with a wide piece of aluminum foil tape as used for HVAC duct joints.

Isolating the feed thrus is seriously non-trivial,good luck.  The people I know with many years of professional TEMPEST experience tell me you almost always encounter weird issues.
 

Online mawyatt

  • Super Contributor
  • ***
  • Posts: 3194
  • Country: us
Re: DIY EMC test cage
« Reply #20 on: September 15, 2022, 05:31:23 pm »
We had all the "Electrical Feedthru" for the mains, Ethernet and Land Phone line (cell phones weren't allowed) handled by a special feedthru fixture specifically designed for such. The mechanical "Feedthru" was handled by other means.

Best,
Curiosity killed the cat, also depleted my wallet!
~Wyatt Labs by Mike~
 

Offline temperance

  • Frequent Contributor
  • **
  • Posts: 420
  • Country: 00
Re: DIY EMC test cage
« Reply #21 on: September 24, 2022, 11:39:34 pm »
I understand the idea of building you're own measurement setup. I've been toying with the idea myself but reverted to a different tactic because TEM cell's and other solutions are too large to handle in a small workplace. (they also take up space when not in use)

You know what your problem looks like if radiation has been measured by an EMC lab. If so, the only thing you need is a way to do relative measurements in a repeatable way.

My take on this has been to place the subject with cables and everything onto a piece of plywood. Fix cables in place with tie wraps and adhesive feet (don't know the name for those in English)

-Measure radiation trough cables with a wide band current transformer. Mark places of interest with sticky tape or something.
-Radiation from the PCB can be investigated with near field probes with the near field probe mounted onto a drill stand with an "extended" (a piece of broom stick) arm. Here you also mark the places of interest with sticky tape.


Primitive but it works.
 

Online nctnicoTopic starter

  • Super Contributor
  • ***
  • Posts: 26752
  • Country: nl
    • NCT Developments
Re: DIY EMC test cage
« Reply #22 on: September 25, 2022, 07:49:43 pm »
I understand the idea of building you're own measurement setup. I've been toying with the idea myself but reverted to a different tactic because TEM cell's and other solutions are too large to handle in a small workplace. (they also take up space when not in use)

You know what your problem looks like if radiation has been measured by an EMC lab. If so, the only thing you need is a way to do relative measurements in a repeatable way.

My take on this has been to place the subject with cables and everything onto a piece of plywood. Fix cables in place with tie wraps and adhesive feet (don't know the name for those in English)

-Measure radiation trough cables with a wide band current transformer. Mark places of interest with sticky tape or something.
-Radiation from the PCB can be investigated with near field probes with the near field probe mounted onto a drill stand with an "extended" (a piece of broom stick) arm. Here you also mark the places of interest with sticky tape.

Primitive but it works.
I get that and I have been using these methods myself for many years but sometimes 'primitive' methods just aren't good enough. The problem I'm hunting appears to consist of not one or two but three seperate problems! The basic problem are emissions from an ethernet cable that also carries power-over-ethernet. It turns out that the PoE injector my customer has been using during the tests by itself is causing enough emissions for a fail. The second problem is that the isolated PoE DC-DC converter on the board and the ethernet PHY are producing junk in the same frequency range. The junk gets onto the cable through the ethernet transformer. However, near field measurements don't cut it because you are always receiving one part of a differential signal stronger. I'm now trying to use a dipole antenna (placed several meters away) with the spectrum analyser set to a narrow band. The downside is that I can only look at very small parts (several 100's of kHz) of the frequency spectrum. Fortunately I have the results from the EMC lab that tell me the exact frequencies to look at.

The test engineer from the EMC lab also brought up using a TEM cell to hunt this particular problem down. I agree the size of a TEM cell with a reasonable useable area is daunting but I see no way around it. For EMC testing you need at least 80cm of useable length to also accomodate the required length of wire. Then again, having a TEM cell could be a tool that is commercially interesting for me because -if it works well- I can offer better pre-compliance testing as a service to customers.
« Last Edit: September 25, 2022, 10:39:48 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline MartinL

  • Regular Contributor
  • *
  • Posts: 51
Re: DIY EMC test cage
« Reply #23 on: September 25, 2022, 11:01:33 pm »
The problem I'm hunting appears to consist of not one or two but three seperate problems! The basic problem are emissions from an ethernet cable that also carries power-over-ethernet. It turns out that the PoE injector my customer has been using during the tests by itself is causing enough emissions for a fail. The second problem is that the isolated PoE DC-DC converter on the board and the ethernet PHY are producing junk in the same frequency range. The junk gets onto the cable through the ethernet transformer.

I have also had problems with PoE devices, they are challenging from an EMC point of view.

However once you know that your emissions are coming from the cable, then there is no need to receive them at a distance with an antenna in order to measure the peaks concerned. You know that you must have common mode RF currents on the cable in order for it to radiate. So a current transformer wrapped around the cable should show you them. A couple dozen turns of magnet wire wound through a clip-on ferrite has done the job well enough in my experience, but you can buy a calibrated RF current probe if you want.

You presumably already have common mode chokes on the TX and RX ethernet pairs, but do you have one on the power side, on the 48V lines where they come into the converter after the diode bridge? A lot of reference designs omit this but I found it made a big difference.

Another big cause of problems in my case was overshoot and ringing on the switch waveform within the isolated flyback converter. A simple RC snubber worked wonders, and some improvements to the converter layout to reduce loop area and improve return paths helped too.

TI app note SLUA469 has some good advice.
 

Online nctnicoTopic starter

  • Super Contributor
  • ***
  • Posts: 26752
  • Country: nl
    • NCT Developments
Re: DIY EMC test cage
« Reply #24 on: September 25, 2022, 11:36:41 pm »
I have a current probe constructed from a clamp-on ferrite with several turns of wire around as well. It works well indeed. I've also constructed a simple two pair ISN.

The problem however is that the current probe output shows a lot of spikes once the ethernet link goes up even with the PoE power supply disabled and the board powered from a clean lab PSU. When I disconnect the ethernet transformer at the PHY side (so the PoE part still works) there is barely any emission coming from the device when measuring at an EMC lab while being powered through PoE. I probably have overdone the filtering in the power side of the PoE part though. Also snubbers, MOSFET switching, real Shottky (without guard band and thus no reverse recovery) have been addressed resulting in great reductions in the near field emissions I measured in my own lab but appearantly also have a positive effect on the far field judging from the results at the EMC lab.

The ethernet transformer does have integrated common-mode chokes but as a test I have added some extra common mode chokes on the board close to the PHY. It makes no difference at all. PHY power supply also looks clean.

But the app note you mentioned (thanks  :-+ ) does show something interesting: Bob-smith termination on the center taps at the ethernet side of the transformer. I don't have that in my design and it is worth at least a try. My current working theory is that there is something wrong around the ethernet transformer. I should not be getting the amount of junk I see from the current transformer when the ethernet link is active. I also tested the previous version from the board which uses the same ethernet transformer but a different PHY and it shows the same amount of junk. So that rules out the ethernet PHY chip.
« Last Edit: September 26, 2022, 12:04:02 am by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf