EMC Chamber build log

[SeanB]

Granite guys here cut the blocks with a small angle grinder with a diamond blade in it. lots of water and grey mess, but it cuts well enough. They use a larger grinder and large slow moving wire saws to cut the blocks into slabs, and a final operation is polishing to a fine gloss finish.

I drilled granite with a masonry drill cooled with water, and it is doable if rather slow. I was not going to polish to a fine finish so cheated with clear matt varnish instead, a lot less work ( about 10 hours of polishing) involved, and a good enough finish over the dull grey it was.

[Tabs]

This is a really interesting read, keep us updated.
How are you doing the ventillation?

Thanks Vgkid,

If you look back at day 4, the last two pictures show the roll of EMC gasket.
The wider field of view shot shows the intake grill, for the ventilation, leaning against the chamber at the bottom near the door frame.
The grill is positioned there. You can see it in its installed position in one of the photos showing Nick's transmit antenna pointing at the door.

This isn't the standard position for the intake grill and I can't remember the reason we moved it to its current location.
If I remember correctly, would normally go higher up on that end piece, I think above the height of the door.

The exhaust grill for ventilation is contained in the ceiling on the diagonally opposite corner. You can see it in the shots of the ceiling going up.

[Tabs]

Why there are ferrite tiles with holes in the middle, while others don't have it ?

Well spotted!. I missed this subtlety.

The center hole is used to screw the board, on which the ferrites are mounted, into the underlying plywood.

I'm not 100% sure on this, but I'm assuming its because the ferrite tile panels (6x6 grid) that carry load (the absorbers) need more screws to take the weight. In our case, we will be using absorbers at the center of each face (including the bottom face). We will use small absorbers so I doubt we need a screw in every tile center. I guess the tile manufacturer can't know before hand what load will be attached to the tiles so gives you a dedicated hole per tile.

It looks like the tile panels that don't have a hole in every tile are only fitted at the edges of each face. I know the absorbers won't cover these areas so only need screwing down in the corners.

I'll ask on Monday when I get in.

[Tabs]

Ferrite is actually very easy to cut on a carbide abrasive saw, like used for bathroom tiles (which are a stronger ceramic).  At a PPoE, we once needed shielding and only had bricks, not tiles or plates; had a shop guy get rather dirty (the black cooling water splashes everywhere..) making slabs.  They were all kinds of rough and uneven, but that didn't matter.  A lapidary saw would be much nicer, if probably slower.

Could probably also cut them like glass or tile, using a carbide scribe to draw straight lines to (hopefully) fracture on.

Tim

and SeanB,

Yes, agree. Not sure why I thought it would be difficult. Using proper tile cutting equipment didn't cross my mind
The water cooling would also help control the dust and smoke which may set off the alarms otherwise.
The installers did have to cut the metal holes for vents and penetration panels outside.
As long as the installers were careful the water wouldn't be too much of a problem.
I think it still makes financial sense not to have to cut the tiles (thus creating a bespoke design). Its just easier on time, transport costs and produces less mess and waste.

[Tabs]

Friday 11th

Foam absorbers starting to go in.
The ones shown in the pictures are the small one. Judging by the size on the pictures these look like 100mm square x 200mm deep.
They are glued to the ferrite tiles.

I believe we have two more sizes of absorbers 400mm and 500mm deep.
I'll confirm once the installers open the boxes.

[lewis]

Looks very cool. Do the tile gaps between the walls and ceiling need to be filled?

[Tabs]

Friday's are half days here,

I took these pictures just as I was leaving.
The ground plane was installed and the bottom face was ready to have ferrites and absorbers placed on them.
This will probably happen over the weekend.

In this picture you can see the 2nd penetration panel that I wanted to get installed. I'm standing were the consumer unit and temporary AC supply is. The far wall is where the antenna is going to be placed.
The primary plan is to have a cable running along the left wall down the full length of the chamber and then across to the antenna.
The EUT will be placed on platform where I am standing.



The 2nd penetration panel is for when I get desperate. I need to squeeze every last Watt out of the 1Ghz amp to generate a 10V/m field and cable losses are playing a big role in this. Above 1Ghz, i need to create 3V/m and this is done by a 2nd lower wattage amplifier. This was done to keep costs down. An amp that can produce 10V/m + 80% Modulation (18V/m) to 1GHz at a 3m separation between the antenna and EUT can easily reach upwards of 30-40k (£). The plan was to re-use our existing AR amplifier and buy a new amp for the 1-3Ghz range.
At the moment I haven't ordered the cables yet, I've been advised by Nick to look at the ultra low loss cables from Rosenburger/Micro-Coax 's UTiFLEX range. Specifically the UFB311A https://assy.micro-coax.com/MCDlgCableDetails.aspx?Cable=UFB311A

I got a quote from Micro-Coax's online store and it costs a little under £1000 ($1400 ish). That'll be the most expensive cable I've ever ordered 

The second picture is taken from where the antenna will be placed. You can see the incomplete section left around the door frame.
When the new door arrives, this area will catch up to the rest of the build.

[Tabs]

Looks very cool. Do the tile gaps between the walls and ceiling need to be filled?

Again, not 100% sure but I'm guessing no. The tiles are required for the low frequency rf range of the test spectrum. The absorbers take over at around 800Mhz to 1Ghz. I don't think the low frequencies will see the gap so won't get a chance to go through it and reflect off metal sheet behind the tiles. Since the absorbers are only being placed in the center of each face, I'm guessing this is where the important area is to absorb the high frequencies and the edges may be less important.

The anechoic tests that Nick will do at the end will tell us if this is a problem. Nick did suggest we could re-use the rectangular absorbers from the older chamber in these areas if we needed to. Again lets wait and see.

[Richard Head]

Strange. All the anechoic chambers I've been in have had the absorbers from floor to ceiling and on the door also. As for the roof I can't actually remember.
I remember seeing a G6 armoured vehicle inside a huge anechoic chamber years ago. It was quite a sight.

[1design]

For the cable you might also want to take a look at the Huber+Suhner Sucoflex 406.

BR,
Miha

[Tabs]

For the cable you might also want to take a look at the Huber+Suhner Sucoflex 406.

BR,
Miha

The full list of cables recommended by Nick were:
Rosenberger Microcoax UFB311A, Huber & Suhner Sucoflex cable, Gore cables or ‘True Blue’ cables

Thanks for the recommendation,
Have you used the SucoFlex?

[Tabs]

Strange. All the anechoic chambers I've been in have had the absorbers from floor to ceiling and on the door also. As for the roof I can't actually remember.
I remember seeing a G6 armoured vehicle inside a huge anechoic chamber years ago. It was quite a sight.

The chamber we are installing will be fully anechoic once complete. This means it will have absorbers on the top and bottom and 4 side walls.
I'm guessing the absorbers in the chambers you've been in were deep which is why you saw them from floor to ceiling, it may not have had tiles behind them. Dave did a EMC chamber facility tour http://www.eevblog.com/2011/09/22/eevblog-202-emc-rf-anechoic-test-facility-tour/ in which no tiles were used in one the larger chambers.
Our chamber is referred to as a compact 3m chamber (Most EMC chamber manufacturers refer to the chamber size by the separation distance between the EUT and antenna). I think the standard calls for 1m separation between the EUT and nearest wall and 1m between the antenna and its nearest wall. Taking these distances into account and including antenna size resulted in a requirement for 7m total length which is just about the biggest size we could fit in the area we selected. This means we have to use shallow pyramidal absorbers backed by ferrite tiles which go from floor to ceiling. Getting bigger pyramidal absorbers would have meant compromises on separation distance.

[Nick Smith]

Hi, I'm Nick - as mentioned by Tabs in some of his posts. I'm new around here and I've been waiting for Tabs to send me a link to this thread, but I gave up waiting and sought it out for myself 

Anyway, I'm happy to answer any questions on anechoic chambers - I've spent most of the past 20 years or so hidden away in one chamber or another and have validated chambers for most of the major chamber suppliers over the years.

I've read the whole thread and would like to pick up on a few of the comments...

Somebody commented about the coverage of the pyramidal absorber not being 100% and that this had differed to other chambers they had seen. The absorber coverage depends on the intended use of the chamber and a chamber for measuring antenna patterns or performing Radar Cross Section measurements would general have full coverage of pyramidal absorber but chambers for EMC measurements are not usually fully covered with pyramidal absorber. Commercial EMC chambers require ferrite tiles in order to get the performance they require down to 30 MHz for commercial testing - unfortunately, putting anything in front of the ferrite tiles will degrade the performance of the tiles and this includes pyramidal absorber.
The pyramidal absorber used in EMC chambers is a Hybrid absorber that is matched to work on top of ferrite tiles - this is done by carefully controlling the carbon doping, however the whole hybrid absorber solution is a bit of a compromise in order to get wider frequency range - the higher frequency performance is not as good as a 'fully loaded' pyramidal absorber that would be used in an antenna chamber and the lower frequency performance is not quite as good as a plane ferrite tile. If you use conventional (non-hybrid) pyramidal absorbers, then the ferrite performance is badly affected.
So the whole absorber layout is a bit of a balancing act and it is generally better to only apply pyramidal absorber where it is most needed. In an EMC chamber at higher frequencies, we generally use log-periodic or horn antenna, which are quite directional - this means that some areas of the room do not require pyramidal absorber, whereas if you want to measure antenna patterns with a 360 degree azimuth rotation, you will need absorber coverage everywhere because at some time during an antenna measurement your antenna will be aimed at the walls.
One final point to note, even for EMC testing not all standards require testing down to 30 MHz and for some (non-commercial) testing, using conventional pyramidal absorber without ferrite is a good solution - the walls in these are then general completely covered in absorber.

Next - holes in the ferrite tiles. As Tabs commented, the tiles have holes so that they can be screwed to the walls. Some manufacturers screw every tile and some pre-glue the tiles to backing boards and then screw the boards (using screws through the tiles at the corners of the boards).
Years ago, it was commonplace to use solid tiles (with no holes) except where holes were required, however after measuring many chambers, it appears that the holes make little or no difference to the final measured result (or not enough difference to make them fail their validation tests anyway). So nowadays, some chamber manufacturers just use tiles with holes. This means they only need to stock one type of tile and they probably get better volume pricing. It also give more flexibility if additional screws are required other than the corner screws. I think they generally put a few more screws on the ceiling panels because it will spoil your day if a 600mm square panel of ferrite tiles falls on you.

Next - The gaps in the tiles between the walls and between the walls and ceiling. These make little difference to the performance (again, not enough to cause a failure on the validation tests). Once upon a time, chamber manufacturers ensured that there were no gaps and that the tiles were basically touching. This requires lots of tile cutting, which is time consuming and messy (it needs a wet diamond wheel tile cutter). The tiles are 100mm square and some bigger chambers may be 18 meters long, 12 meters wide and 6 meters high (or even larger) - that's a lot of tile cutting if you have to cut into every corner!
There is a limit to the size of gap that you can get away with, but 20 or 30mm generally makes no significant difference so nowadays they seem to try and minimise the amount of tile-cutting required.

The foil backing behind the tiles is not related to the wood being hygroscopic as Tabs suggested. The (RF) wave is attenuated as it passes through the tile, it then reflects and is attenuated as it returns through the tile. The spacing between the tile and the reflector affects the phase relationship between the direct and reflected waves and this consequently changes the performance (due to constructive or destructive interference of the two waves). Obviously this is frequency dependant, but If the spacing between the tile and the reflector is too large, then you end up with a chamber that doesn't perform well at some frequencies. Without the foil, the reflector would have been the shield itself and with this type of shield construction, the shield is spaced 60 or 70 mm away from the tile, so the foil is used so that the dielectric spacing between the tile and the reflector can be controlled. This spacing is actually quite critical and can be used to 'tune' the broadband performance of the tile. Traditionally, this was used to improve the higher frequency performance of the ferrites so that these rooms could be used to 1 GHz. Nowadays, the hybrid absorber takes over from the ferrites around 700 or 800 MHz and the fine tuning of the tiles is less important, however if the dielectric spacing is too large then you will not get the required performance from the tile.

I think that's about it from me for now - I've written a bit more than I intended too...
I'm happy to answer any other questions and if anybody has any shielding effectiveness measurement requirements or needs an anechoic chamber measuring, then please get in touch!
(I'm not sure whether it's against forum rules here to post a link to my company's website so I won't. Tabs might post a link though...)

[Tabs]

Hi, I'm Nick - as mentioned by Tabs in some of his posts. I'm new around here and I've been waiting for Tabs to send me a link to this thread, but I gave up waiting and sought it out for myself 

......

The foil backing behind the tiles is not related to the wood being hygroscopic as Tabs suggested. The (RF) wave is attenuated as it passes through the tile, it then reflects and is attenuated as it returns through the tile. The spacing between the tile and the reflector affects the phase relationship between the direct and reflected waves and this consequently changes the performance (due to constructive or destructive interference of the two waves). Obviously this is frequency dependant, but If the spacing between the tile and the reflector is too large, then you end up with a chamber that doesn't perform well at some frequencies. Without the foil, the reflector would have been the shield itself and with this type of shield construction, the shield is spaced 60 or 70 mm away from the tile, so the foil is used so that the dielectric spacing between the tile and the reflector can be controlled. This spacing is actually quite critical and can be used to 'tune' the broadband performance of the tile. Traditionally, this was used to improve the higher frequency performance of the ferrites so that these rooms could be used to 1 GHz. Nowadays, the hybrid absorber takes over from the ferrites around 700 or 800 MHz and the fine tuning of the tiles is less important, however if the dielectric spacing is too large then you will not get the required performance from the tile.

I think that's about it from me for now - I've written a bit more than I intended too...
I'm happy to answer any other questions and if anybody has any shielding effectiveness measurement requirements or needs an anechoic chamber measuring, then please get in touch!
(I'm not sure whether it's against forum rules here to post a link to my company's website so I won't. Tabs might post a link though...)

Hi Nick,

Welcome to the EEVBlog forums.

Apologies for not sending out the link earlier.

Thanks for the input & corrections.

Here's a link to Nicks company website:
http://www.wave-scientific.com/

I think its ok for the link to go in the thread.
If Dave or an admin takes issue with it please let me know and i'll modify the post.

[Tabs]

Wednesday 16th,

New shorter door arrives and is installed. The ground plane near the door is ripped up to remove the old door frame. You can see a small 100mm blanking panel at the top.

[Tabs]

Plywood fitted around door area and foil lining glued into place.
500mm absorber being fitted to main wall
main wall complete and ferrite tiles added around door frame.

[SeanB]

At least you did not do as I saw in one building and cut the beam away. That was done so the forklift could pass under it, and they were stacking on the floor above as well with crates 2m high. I was always wondering when I was there if the floor would decide that that was the day it was tired.

[nctnico]

Moral of the story, if you don't fully lock the chamber door or start putting wires holes not meant to pass wires then you may as well have the door fully open, or no chamber at all. (at least at that test frequency).

Putting ferrites onto a cable going into the chamber works very well to keep unwanted signals out. A couple of months ago I attended some EMC tests. We got a spike at 90MHz I couldn't explain. It turned out the tech forgot to put the filter on a cable going into the chamber.

[lpc32]

Interesting. What's the temperature in there?

BTW, very artistic/mysterious those glare/blur photos.

[Tabs]

Moral of the story, if you don't fully lock the chamber door or start putting wires holes not meant to pass wires then you may as well have the door fully open, or no chamber at all. (at least at that test frequency).

Putting ferrites onto a cable going into the chamber works very well to keep unwanted signals out. A couple of months ago I attended some EMC tests. We got a spike at 90MHz I couldn't explain. It turned out the tech forgot to put the filter on a cable going into the chamber.

I'm not sure what your'e describing nctnico. I think your describing a penetration panel designed to accept your products cables and which allows you to monitor/control the product/EUT. These signals are usually filtered and the chamber boundary and any screens also attach here. If you tech forgot to add the filters in this setup then I think you resolved it correctly.

What I was describing was when people make use of vent holes or the waveguide to pass cables through the chamber boundary when the penetration panel doesn't have the interface they need.

Although you are right (in principle) that you can use a ferrite clamp on a cable to resolve problems like what you describe in the setup above, I wouldn't recommend it for general use when you don't have a compatible penetration panel. For me, the difficulty would be showing the 90Mhz spike was from the EUT (using an accepted EMC standard) and not and external signal being coupled into the chamber along the wire. The antenna wouldn't be able to differentiate between external source or EUT. At worst you may have added a ferrite clamp to your BOM for the product cable when in fact it wasn't needed. If the 90Mhz was from the EUT it's better to find the source (using something to act as a field probe) and fix it (time/money dependent).
Its just something that I would say causes more headaches for keeping traceable records and proving your EUT complies with EMC regs. I know in europe, if you self declare with the technical file approach, you need to keep the file on hand for 10 years from the point the product was last placed on the market.

[Tabs]

Interesting. What's the temperature in there?

BTW, very artistic/mysterious those glare/blur photos.

The chamber is located down a corridor from the main product area where we have a lot of wave soldering machines. I'm told the temperatures used to get very warm in the summer but I haven't experienced the problem because i've not been there long. Since my arrival, we have had new air-conditioning installed. Its a heat exchange system which takes heat from our reflow ovens  in SM and distributes it through the building in winter or takes heat out of the building in the summer.

The chamber should be at what ever the ambient temperature of the environment is. There are no heat sources radiating onto the chamber and most of our products are less than 15-20W so there shouldn't be any heat building up inside.

[Tabs]

Extra artistic/mysterious glare/blur photo for Ipc32.



Thursday 17th
2nd pic - electrical conduit and sockets being routed and made ready for electrician to run his cables through
There will be sockets at the front (for EUT) and at the back where the antenna is (for camera), light switch near door and an emergency light above the door.



3rd pic - white polystyrene tiles are glued into shape and fitted on top of blue pyramidal absorbers.

[Tabs]

[lpc32]

Extra artistic/mysterious glare/blur photo for Ipc32.

Oh my god, it's full of glares!

BTW, not bad for a cellphone camera.

[LapTop006]

...white polystyrene tiles are glued into shape and fitted on top of blue pyramidal absorbers. ...

Are those just impact protection, or do they actually serve an RF purpose?