EMC Chamber build log

[Tabs]

Hi all,


We recently had to upgrade our in house EMC chamber for pre-compliance testing of our products.
The old one was too old (20 years ish) and didn't have the structural support to be upgraded to modern specs.
The chamber was teared down and scrapped and our mechanical workshop was cleared to make way for the new (bigger) chamber.

I was supposed to do a daily build log of the chamber but other work commitments made it difficult to upload here until today.
I was doing work stuff at home to meet a deadline for an exhibition in China. This made sleep my next highest priority 

Anyway, I've built up a collection of pictures that will have to be uploaded in separate posts to this thread due to the attachment limits.

If anyone would like to know more about whats involved in the process of building an EMC chamber please feel free to ask any questions. I'll answer as many as I can and forward those that I can't to the two EMC companies I hired to provide the chamber and independent testing of the chamber. The EMC consultant has offered to come onto the forums and answer questions directly. You should definitely use the opportunity to pick his brains about EMC.

Here are the first few pictures:

 

Mechanical workshop cleared (nearly), external windows & doors removed, external wall (far side) bricked up, near walls (internal partitions from where I took the pictures) are removed to allow room for chamber material.



2 months later, EMC chamber material arrives. Area fully cleared and environmental membrane layer laid down.

[Tabs]

Chamber floor being bolted together. I've not measured the individual panels but just estimating that the larger ones measure 2mx1m and the smaller ones 1mx1m. Each probably has a 30-40mm folded lip at the boudary. I'm assuming 4-5mm thick which makes each panel somewhere between 75-100Kg (165-220lb). There are two guys doing the build.

[Tabs]

Day 3

Side walls going up. 



EMC gasket being persuaded to go in the right position. Panel being bolted through.



Next panel being made ready



Chamber installers on a break.
View from the inside looking towards where I normally stand for the pictures. You can see the top left area where the chamber door is going to be.

[Tabs]

Inside view of back wall

[Tabs]

Day 4

Final layer of side walls fitted


Inside view looking at back wall (where the antennae are going to be positioned)


Inside view looking at front of chamber where the EUT will sit. door opening is on the left of image.


Structural steel supports in place ready for ceiling. You can also see the hooks that the knife lock on the door will engage with.
As the door handle moves towards is closed position the door is compressed against the chamber wall and EMC seals. More on how this affects EMC shielding performance later.

[Tabs]

Closer look at the EMC gasket for those who are interested.

[Tabs]

Day 5,

Ceiling panels begin going up.


Ventilation panel fitted (extracting air)


ahh good ol 2x4 doing its job 

[Tabs]

Ceiling complete.




Door attached.


Penetration panel hole cut and fitted. You can also see a black box attached to the side of the chamber. This is a 30A filter to give us a clean supply inside the chamber. The unit takes 230VAC mains and 24VAC & DC low voltage mains. There is a waveguide for optical fibre cable which we use to for isolated comms to the EUT inside the chamber.

Phase 1 complete. Ready for the shielding effectiveness test.

You can see the transmit horn antenna directing a signal at the penetration panel.
The chamber is being powered off an extension cable (real supply not installed yet). This is used to provide a clean supply to the  receiver (spectrum analyser) on the inside of the chamber when the door is closed.

Once the chamber is complete, we intend to rebuild an internal wall (partition) just where the left side of the ladder is. The wall extends out to where I am standing in the last picture. The area between the partition and the chamber will be where the control  equipment and amplifier rack live. Nice and cozy )

On the the other side of the wall is the corridor leading to our surface mount and reflow soldering ovens left and the canteen on the right.

[T3sl4co1l]

Cool!

Can't imagine that pallet of ferrite tiles (upcoming I hope??) is very cheap!

Tim

[Tabs]

Cool!

Can't imagine that pallet of ferrite tiles (upcoming I hope??) is very cheap!

Tim

I think I've got another 2 or 3 days worth of pics to go through yet before I get to the ferrite tiles.
Your'e right about the cost of the ferrites. They can be more than 50% of the entire chamber. I managed to get a really good deal but let me make sure with my boss before I start talking about the money side of things.

[Tabs]

Shielding effectiveness testing begins.

Different antennae are used for certain frequencies in both vertical and horizontal polarisation.

The transmit antenna is directed at the receive antenna and both are separated by a set distance.
This setup is used take a measurement without shielding.

Unfortunately, I cant find the picture I took so just imaging two horn antennas pointed at each other.

The receive path is then setup inside the chamber and a measurement is made with the door sealed.
Subtract the two measurements and you get shielding effectiveness of the chamber.

[Tabs]

Here's Nick, our EMC consultant who's been hired to take independent measurements for us. He can be seen checking the shielding around the door frame.



Here's the transmit signal with the two antennas pointed towards each other outside the chamber.



You can see the difference in signal strength when the receive antenna is moved inside the chamber and the door closed.


We started to experiment with the shielding and Nick showed us what happens when you don't fully lock the chamber door.
The knife lock was left about 2-3mm away from its locked position, you couldn't even fit your finger nails in the gap created between the door and the chamber wall. The result was the same as when the receive antenna was pointed directly at the transmit antenna.

We then poked a length of 7/0.2 wire through a hole in the ventilation panel. The result was the same as having the door open.
Nick was telling me how some people get desperate enough to do this or poke wires through the waveguide tube which is meant for optical fibre cables. This usually happens when the penetration panel doesn't have enough connection points.

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).

When installing a chamber, make sure you think carefully about the requirements of the penetration panel. Otherwise you will have to swap panels and this could change the shielding effectiveness.

[G0HZU]

Where I work we used to have our own EMC Compliance subdivision with various EMC chambers of different size and capability. eg tiled and/or pyramid RAM. I've been in loads of EMC/anechoic chambers over the years so it's interesting to see how a modern one gets built.

Are you going to tile the chamber and also fit some pyramid RAM?

One comment I have about your new chamber is that the door locking mechanism looks very simple and light. Maybe this is down to modern/better design but the chambers we had at work had huge door latching mechanisms that looked like something from the back of a huge door mechanism from a truck. i.e. you had to use your body weight to move the mechanism and drag the heavy door open.

Note: Probably the biggest/most impressive anechoic chamber I visited and worked in was the one at the McAfee Centre/building 600 at Fort Monmouth in New Jersey USA in around 2005. It was huge!

[G0HZU]

We started to experiment with the shielding and Nick showed us what happens when you don't fully lock the chamber door.
The knife lock was left about 2-3mm away from its locked position, you couldn't even fit your finger nails in the gap created between the door and the chamber wall. The result was the same as when the receive antenna was pointed directly at the transmit antenna.

I wondered if it would be possible to also use ultrasonics to test the new door seal... Many years ago I knew someone who used to sell a simple and low cost seal testing system (for cars or fridges etc)
The idea was to put a high power ultrasonic source inside the car or fridge or room and then run a tuned/sensitive detector all along each seam. It could find the slightest leak in the seal within about 2cm as the sensor sniffed along the door or window seals. This is because the ultrasonic energy can leak out of the seal.

I know this isn't really testing it at RF but it's a very good method to find 'real' faults in a door seal

[Tabs]

We started to experiment with the shielding and Nick showed us what happens when you don't fully lock the chamber door.
The knife lock was left about 2-3mm away from its locked position, you couldn't even fit your finger nails in the gap created between the door and the chamber wall. The result was the same as when the receive antenna was pointed directly at the transmit antenna.

I wondered if it would be possible to also use ultrasonics to test the new door seal... Many years ago I knew someone who used to sell a simple and low cost seal testing system (for cars or fridges etc)
The idea was to put a high power ultrasonic source inside the car or fridge or room and then run a tuned/sensitive detector all along each seam. It could find the slightest leak in the seal within about 2cm as the sensor sniffed along the door or window seals. This is because the ultrasonic energy can leak out of the seal.

I know this isn't really testing it at RF but it's a very good method to find 'real' faults in a door seal

I've heard of people using ultrasonics to test the integrity of materials (stress, strain, micro-fractures etc) and the integrity of things like welding joints. I'm no expert on it but I guess you could use it to find 'real' faults.

The only thing is that it probably only tells you the mechanical nature of the fault and not the impact on the electrical performance or parameters important for an EMC chamber. An example is:
it would tell you there is a gap between two panels but it wouldn't show you the effect of the gap on RF radio waves at different frequencies. You would see variations across the spectrum and then huge drops in the shielding effectiveness as the frequency approaches a resonant peak formed by the the chamber & gap dimensions. This probably means you couldn't use ultrasonics to get a full picture of the chamber performance.

One of the older EMC experts that I used to work with at my old job told me of how gaps between panels due to EMC gasket failures created electric field strengths inside the chamber that were bigger than the field being produced outside the chamber. This was all down to resonances inside the chamber.

BTW - that can catch some people out, if you have a metal enclosure around your product and go to EMC test expecting it to be hit with 10V/m (for example), you could actually expose the electronics to bigger fields inside your enclosure.
I used to have paper that described the problem.
If anyone's interested I'll dig it out.

[lewis]

Very interesting stuff, I'll be following 

[Tabs]

Where I work we used to have our own EMC Compliance subdivision with various EMC chambers of different size and capability. eg tiled and/or pyramid RAM. I've been in loads of EMC/anechoic chambers over the years so it's interesting to see how a modern one gets built.

Are you going to tile the chamber and also fit some pyramid RAM?

One comment I have about your new chamber is that the door locking mechanism looks very simple and light. Maybe this is down to modern/better design but the chambers we had at work had huge door latching mechanisms that looked like something from the back of a huge door mechanism from a truck. i.e. you had to use your body weight to move the mechanism and drag the heavy door open.

Note: Probably the biggest/most impressive anechoic chamber I visited and worked in was the one at the McAfee Centre/building 600 at Fort Monmouth in New Jersey USA in around 2005. It was huge!

The walls and ceiling will have 100% coverage of ferrite tiles. On top of the tiles we will have the traditional pyramidal RAM placed in the center of each wall, ceiling and floor. On top of the RAM we are getting what looks like white polystyrene tiles. I'm not sure if these are RAM also, I'll ask.

If you look closely, you can see the door is too large and collides with the low hanging wall. I'll talk more about this later, but this is going to be resolved sometime next week with a new smaller door.

The standard door does take some effort to open, but not body weight amounts, (not my body weight anyway).

[Tabs]

Day 6 (Weekend - Saturday 5th April)

Most high end chambers achieve shielding effectiveness around 120dB or more.
Placing too high a spec on this requirement can drive up the price of the chamber. This is especially true if you want to maintain that level of shielding up 18Ghz or so; which is what you need if you do military EMC testing to MIL-STD-461, DEFSTAN 59-411 or STANAG.

Most commercial products don't need that level of performance from an EMC chamber. In our case we test to EN61326 for EU and CISPR 11 for FCC when we sell into the American market. I went with an 80dB shielding spec to 3Ghz after consulting with Nick, the chamber providers and some commercial test houses were we know helpful people. 3Ghz is the max we need to test at the moment.
The chamber is capable of going up to 18Ghz.

This was one of the ways we were able to keep the cost of the chamber down. Shielding effectiveness at this stage is really down to how well you tighten the bolts and the EMC gasket performance. The expectation was that it would be very easy to achieve better performance than the requirement I set.

In a few years, we may be required to test to 6Ghz due to newer mobile phone and wifi standards. But these requirements will take a long time to supersede the current standards and it was cheaper to purchase equipment up to 3Ghz and wait to see what happens.

The shielding tests were completed and only two problems were located:
1) missing EMC gasket between two panels
2) a single nut & bolt not tightened to the right settings to give the required sealing.

Nick only had access to two faces of the chamber. The 1st was direct line of sight access to the control area where the penetration panel is. The second was through a breeze block wall in the room next to the chamber.

The rf signal passes through the wall and hits the chamber side wall. The signal travels around the chamber since its acting like a Faraday cage. If there are any gaps in the shielding the rf leaks into the chamber where Nick was able to detect it.

Both problems were fixed easily.

The final measurement for shielding effectiveness came to 110-115dB. Nick said if he brought his higher end pre-amp we could probably squeeze a few more dBs out of the chamber. At this point we've pretty much got a pre-compliant chamber with shielding effectiveness on par with full blown test houses.

[Tabs]

Some of you may have noticed from the later pictures that the door is too tall for the opening in which it was meant to swing.
The original intention was for the door to swing under the low hanging wall. I measured a good 30mm clearance between what I thought was the top of the door and the bottom of the low hanging wall. I hadn't realised the door was raised by 100mm because it wasn't indicated on the plans.

A perfect combination of assumptions and misinterpretation. Shit happens.

The door was only able to open through an arc of 400mm before it collided with the wall and this meant it would be difficult to transport the material for the next stage of the build into the chamber.

It also had the potential to delay the schedule by 3 weeks because the shielding test couldn't be completed without the door.
Its always better to get the shielding test done before the next stage starts. Can you imagine fitting all the ferrite tiles, absorbers, wiring etc and then do the test only to find out that 1 bolt wasn't tight enough and you missed an EMC gasket 

We decided to complete the chamber with the larger door so that Nick could come in test the chamber. The idea was that he could find problems in other areas that we could fix then and there. After the testing, the large door was going to come off and the next stage would start. All the material could now be moved in. The area around the door frame would be left until the last possible minute. This allowed the 2nd stage to stay on track whilst a shorter door and blanking panel were made up and delivered.

The solution was proposed by the chamber installers who are contracted by the company who supply the chamber. They are on track to deliver the new door mid next week and at no charge to my company. All very professionally handled.
If you need a chamber check out http://www.globalemc.co.uk/ highly recommended.

Anyway, I arrived back to work the following Monday to this:





2x4 batons were laid along the floor, ceiling and side walls. 10mm plywood was screwed into the batons on the floor. The chamber smelt of fresh sawdust.

[Tabs]

More plywood going on.

The wood is hydroscopic so its water content changes throughout the day with ambient environmental conditions like humidity and temperature. Not the best thing for creating uniform field strengths inside the chamber.

Hence the aluminium foil lining glued on top of the plywood.

[Tabs]

Tuesday 8th April,

Aluminium lining has been glued to the inside surfaces and ferrite tiles are going up 

When its all complete, 1.5 tonnes of ferrite tiles will cover every inch of the inside.







As you can see from the last picture, there are two types of tiles going on.
I'm told the dark gray tiles are Samwha tiles. These are the high quality tiles. After a quick google search I think I have these ones:
SFA600 or SFA600A
http://www.samwha.co.kr/SW_catalogue/catImage/37/Ferrite_Absorber.pdf
Each individual tile is 100mm sqaure. They are mounted on a board in a 6x6 grid.
I'm told each tile costs $5 US direct from the manufacturer,  don't know how much we paid for each.
I worked out (roughly) that we would need a little over 10,000 tiles to fill the inside of the chamber so $50k just on tiles would not leave me with anything for the equipment.
Global EMC offered some 2nd hand tiles (which are hard to come by; its not like people scrap a ferrite lined chamber on a daily basis). These are the light gray tiles in the photo above.
I'm told they are some sort of Phillips tiles but not sure which specific type.

The good tiles are going to be placed strategically on the important walls and then less strategically on the side walls. The Phillips tiles will then cover the remaining area on each surface. Using the 2nd hand tiles let us keep the cost down.
When Nick comes in again to do the anechoic tests, we will see if this payed off.

[Tabs]

Thursday 10th

Installation of ferrite tiles complete (except the area around the door frame)
You can see the ferrites are practically edge to edge.
The chamber dimensions are designed around the dimensions of the ferrite tiles. This means the chamber size can only change in increments of 100mm. This guarantees edge to edge coverage of ferrite tiles without having to cut them. These things are very brittle so cutting them would be next to impossible and probably result in a lot of wastage.
This is another way to keep the costs down.

In the final photo, you can see the area where the penetration panel and consumer unit sit. There's a temporary socket placed inside which was used by Nick during the shielding tests.

[Vgkid]

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

[BravoV]

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

[T3sl4co1l]

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