1 Building your own TEM cellIntroduction![](https://i.ibb.co/xzC4RkK/20230115-211731.jpg)
One of the problems I'm facing is that EMC testing can become quite complicated for some products I develop for customers. The entire test setup can influence the test results and wiring plays a large role. Especially looking for particular frequencies that are slightly over the limit is difficult because measurements done outside a shielded chamber are easely drowned by noise that is floating around in a typical office space. Spending a lot of time at a test lab is also not ideal and expensive. Customers like to keep the number of hours spend at a test lab to a minimum while I would like to do a bit more optimisation on the design. So I wanted to have more possibilities to do more & better EMC testing in my own lab.
Shielded chambers are very expensive and I don't have the anywhere near the space for something like that. I tried to build a small Faraday cage (80cm x 100cm x 60cm) for this purpose using chicken fence (see:
https://www.eevblog.com/forum/rf-microwave/diy-emc-test-cage/msg4394122/#msg4394122 ) which worked great for keeping the noise out in the frequency range I was most interested in at that momment. But it didn't really work well for doing measurements that resemble the far field measurements done in an EMC lab. For a particular product I had to resort to doing tests using an antenna a couple of meters away from the DUT in my office. With the help of a pre-amplifier and very narrow bandwidths I managed to capture some of the offending frequencies on my spectrum analyser and see what caused them. But getting a good grip on what was noise and what was an actual signal was hard.
So I wanted something else that would work better than the small Faraday cage. One of the options would be to get a large tent made from conductive cloth. IOW: A big Faraday cage. These can be bought commercially but the fabric is also available from Aliexpress and so on. Add some PVC or aluminium piping as a frame and done. Still, such a tent would be huge and it may cause reflections.
So I looked further and found a device called a TEM cell which is useful for EMC testing (
https://en.wikipedia.org/wiki/TEM_cell. The CISPR25 standard even has some guidelines on the relative sizes of a TEM cell and what part of the TEM cell is actually useful for doing measurements. One of the technicians at the EMC lab also suggested using a TEM cell as a means to have a 'compact' testing facility.
A TEM cell is essentially a rectangular metal chamber with a flat, plate shaped conductor (called septum) in the middle that can be used for doing radiated emission testing and to expose DUTs to very high electric fields. I like to think a TEM cell works sort of like a coax cable. One of the properties is that the electric field is homogenous in a significant part of the chamber parallel to the septum. An added bonus is that a TEM cell can also be used to test immunity very efficiently. Very little power is needed to create large electric fields. All in all I think/hope having a TEM cell would allow me to offer more services to customers where it comes to pre-compliance testing and hunting EMC problems down. Reasons enough to spend some time & money on acquiring a TEM cell.
Now how to get one cheap? Or maybe take a step back first: what do I really need? My primary goal is to be able to do pre-compliance testing for CE and FCC where it comes to emissions. CE requires 80cm of cable and FCC even 100cm to be included in the measurement setup (if the device has a cable which is usually the case). So I would need a TEM cell that can accomodate a test setup of about 1 meter long. I also wanted to have a reasonable space for the DUT. The field is homogenous in about 1/6 of the total height. A useful height in the ballpark of 8cm would be nice. I went looking to see what can be bought but I only found rather small TEM cells from Tekbox which where open at the sides (which doesn't solve the issue with external noise). Bigger ones are quite expensive. So I decided to build one myself.
Reading materialsDesigning & building something starts by getting an idea on what is involved and how to construct something. There is lots of material that can be found but IMHO these are the most interesting documents. I have based my design on formulas, information and hints found in these documents:
The most useful information is a document by Crawford (more or less the inventor of the TEM cell) with lots of emperical results and formulas to calculate the impedance of a TEM cell:
Using a tem cell for emc measurements
https://www.govinfo.gov/app/details/GOVPUB-C13-6e0dabf32dffece069e3fea4415bca89 (reference 1)
And a deeper dive into calculating the characteristic impedance of a TEM cell (reference 5 from the document above):
https://www.govinfo.gov/app/details/GOVPUB-C13-6c23cf9d22f3b1f99517c8891a966a50 (reference 2)
Since the early 1970's many people have worked to improved the idea. Modern day simulation tools help a lot to tweak a design without actually having to build it. A few useful documents below:
https://www.researchgate.net/publication/307167794_TEM-Cell_With_Increased_Usable_Test_Area (reference 3)
https://www.jpier.org/ac_api/download.php?id=21081501 (reference 4)
Based on this information I started to work on a design.
DesignLet's set some goals first. Most of the EMC problems I've seen so far are at frequencies of a couple of hundred MHz at most. The most common switching power supplies and DC-DC converter modules create problems below 100MHz. So I don't need a TEM cell that works up to several GHz. Up to 1GHz should do it. So the goal is to make a 50 Ohm TEM cell that is usefull to about 1GHz and can accomodate a setup of about 100cm long and 8cm high.
The impedance of a TEM cell is driven by the width and height. A normal TEM cell has the septum (central conducting plate) in the middle but I decided to offset it a little bit (30-ish %) to create more room for the DUT without making the TEM cell much bigger. This means that the electric field is less homogenous though. A downside of a bigger TEM cell -besides storage space needed- is that the resonant frequencies also become lower which limits the upper frequency where the TEM cell is useable. All in all I have traded a less homogenous field for a smaller TEM cell with a higher working frequency.
Because I started to drown in the formulas quickly while iterating through what would be the ideal size, I decided to create a spreadsheet to calculate the various parameters based on a few simple input values (width, height, length, sheet size, tapered section angle, size of the square opening, etc). See the attached Excel sheet. The Excel sheet basically outputs all the mechanical sizes of a TEM cell based on a few simple input parameters. Based on the angles from the document by Crawford et al, I decided to have 45 degree angles for the top and bottom plates that form the pyramid shape. The angles of the sides are calculated from the various input. However, the impedance calculation is not correct for TEM cells with an asymmetric septum. See the 'Measurement & tweaking' chapter.
I already noted that a TEM cell can also be used to do immunity testing. The Excel sheet also calculates the required power level to achieve a certain electric field strength. If the calculation is right, then 0.11W (20dBm) should be enough to get to 10V/m. I have done some immunity testing at a lab and the amplifier over there had to blast 25W into the shielded room in order to expose the DUT to 10V/m.
Since the shape is much like an airduct found in ventilation systems, I looked whether such a duct could be an option but I didn't find anything useful. So I choose to construct my TEM cell from scratch using 100cm x 50cm 1mm thick aluminium sheets that I can buy from a store nearby. I divided the TEM cell into 3 sections: 2 end pieces (pyramids + rectangular tube section) and a mid section (rectangular tube) which I can bolt together. I used the size of the aluminium sheets determine the width of the TEM cell and the length of the two end pieces in order to reduce cutting losses. I could have ordered larger sheets but then I'd need pay extra delivery costs and also would need to cut these myself. Based on the sheet sizes and dimensions I found in the various documents, I ended up with a height of 36.2cm (inner height of 34cm), a 50cm width (width of a sheet) and a total length of 1.75 meters
I also needed to consider the limitations I have where it comes to tooling and size of my work space. Ideally it would be nice to bend 90 degree flaps at the edges of the sheets but I'd need a larger brake for that than I have space for. So I decided to use readily made 10x10mm L profiles to make flanges that hold the sheets together using M3 bolts and rivets. This also eliminated needing to bend sheets very precisely in order to have matching bottom / top plates.
And then there is the septum. When you look closely at the septums used in the small TEM cells sold by Tekbox, you'll notice those are made from circuit board material and they are not continuous conductors but copper strips. This is done to supress resonances; this is also explained in the 3rd PDF I linked to above. Based on the septum sizes from the Excel sheet, I created 2 PCB designs: 1 for the tapering section and 1 for the straight part. I decided to cut the straight part into 4 boards. Otherwise the PCB would become too expensive to produce. JLCPCB still had a good day though when I send the order in. I think these are the largest PCB boards I ever designed. I didn't go into optimising the tapered section for the best impedance matching.
Last but not least a TEM cell needs some kind of door and a way to get signals in. Since the mid section is 100cm long, I decided to put the door in there like a door on an oven. I want access to both lower and upper sections so I ended up with a door design that is 80cm wide and 21cm high. Ofcourse this door needs to be RF tight so the design needs to accomodate the use of EMC fingers. Another job is getting signals in. I wanted to keep this part sensible and not make it too versatile. I decided to use 3 BNC bulkhead connectors, 4 SMA bulkhead connectors, a filtered sub-D 9 bulkhead connector and 2 keystone module holders. Typically you fit the keystone holders with modular jack keystones, but you can also get modules with USB-A and USB-C connectors. This gives me great flexibility where it comes to testing ethernet and USB powered gdgets. I specifically avoided banana-plugs. No banana-drama! Instead, the filtered sub-D 9 bulkhead is intended to feed auxilary signals through that don't work well with BNC, SMA or keystone.
2 ConstructionThis is a long winded chapter about tooling, cutting, bending, drilling, rivetting, filing, etc aluminium. Unfortunately it was full on winter until I got to the actual construction so I nearly froze my fingers off while working on the TEM cell.
ToolingFor this job I needed to be able to deal with aluminium sheet and aluminium profile. I decided to cut as much as possible using a circular saw. The advantage is a super clean cut and sheet metal stays absolutely flat. For cutting aluminium with a circular saw you need a blade with trapezium teeth; I had to get another one for my plunge cut saw in order to cut flat sheet but these saw blades aren't too expense.
Ready for the first cut:
![](https://i.ibb.co/wSGpp3j/20221126-152408.jpg)
Then I still needed to cut a few pieces using a metal shear. The blades on my good old Paddinghaus metal shear are not that good and the blades are out of alignment. A relative still had a set of new blades but these didn't fit the model I have. I quickly learned that buying a different Paddinghaus shear that matches the blades was cheaper than the right blades for the meatal shear I already had. Fortunately one was for sale nearby. After changing the blades and shimming the lower blade to make it align with the top blade, I have a nice metal shear with sharp blades.
Making the rectangular hole for the door in the TEM cell concerned me a bit. I could make it from pieces of sheet pieced together but I wasn't to keen on doing that. I'm afraid that it would be too weak. Another relative suggested to use a plasmacutter. A plasmacutter looked like a nice tool to have so I got one. Only to realise that I don't have enough electric power in my shed to run both a compressor and the plasmacutter. This ended up with adding a 25A circuit (GFI / breaker combo) to the distribution panel inside the house, pulling a 6mm^2 cable under the house and through the garden into my 2nd shed. A small distribution box for two 16A circuits sits at the end of the cable with 1 outlet. I still need to tidy this up by digging a ditch to put the cable under the ground properly and feed it into the main shed properly. The plasma cutter is nice to have though; it also allows me to cut steel in an easy way. Until now I didn't really have a tool to cut steel quickly. I already used it to create a lamp holder on my milling machine. But only after I bought a couple of safety goggles (3M, locally sourced; not from Amazon/Aliexpress) to protect my eyes from UV radiation coming from the torch. A steel grid serves as a table top for plasmacutting.
![](https://i.ibb.co/vDLmXsz/20221211-135446.jpg)
![](https://i.ibb.co/TMHMdtd/20221211-141437.jpg)
Next on the shopping list is a brake to bend sheet metal. After lots of pondering (see
https://www.eevblog.com/forum/mechanical-engineering/which-(press)-brake-for-sheet-metal-folding/) I decided to spend some money on a more elaborate Chinese brake from a local webshop. The good quality second hand ones I found where too big and too expensive. I wanted to make sure I ended up with nice, straight bends so the pyramid shapes sections line up nicely.
Ofcourse I need to drill lots of holes as well. For that I want to use my good old drill press. And when I need to drill a lot of holes, I like to do that at high RPM. Only thing was that my drill press really needed an overhaul after 30+ years of service... So before commencing with the TEM cell project I took it apart, cleaned the parts and replaced the bearings. All the bearings where dry and several where worn out. SKF had a good day this time. Also the spindle needed lots of new grease. Even though it is a relatively cheap Chinese drill press, I must say that I'm surprised that the mechanical parts (axles, spindle, spindle housing, etc) that really matter where not rusty at all. Despite the fact that it has been sitting in a damp shed all the time. And I redid the wiring as well. It looked quite dodgy... Maybe I'll change out the motor some day as well. The bearings in the motor don't sound that healthy and I could use a motor that is slightly more powerfull.
Assembling techniquesOne of the things I wasn't sure about was how to attach the aluminium L profile to the sheet. Soldering, rivetting or both? Soldering would provide a good electrical connection so this seemed worthy to investigate further. You can buy cheap aluminium soldering rods from Aliexpress but after watching some reviews, I decided to buy soldering rods from a reputable brand. I could source Bernzomatic locally so I bought those. First I tried my soldering iron and hot air (to 600 deg. C) but that didn't work. I needed to use my butane torch to get the material hot enough in order to melt the rod. The rod seems to be some kind of zinc/copper/aluminium alloy. It is quite hard, definitely not soft like aluminium. I think there is also some fluxing agent in there. It takes a bit of practise and sometimes you end up with an ugly lump of solder on your workpiece that just doesn't want to stick to anything. Just like lead/tin solder does when the flux has burned off. The lump needs to be wiped away and applying fresh solder makes the joint flow much better. I've noticed it is really important to heat up the workpiece and not the soldering rod to get the best results. Reading various reviews for these kind of aluminium soldering rods, I think some people don't do that and get poor results.
I used these but the ones from Harbour Freight (which I picked up during a trip to the US) work equally well.
![](https://i.ibb.co/MZWV1TG/20221217-173314.jpg)
Back to soldering the L profile to the sheet: that doesn't work well. I made a test piece and the sheet gets warped quite a bit so I decided to use only rivets for attaching sheet the aluminium profile. I did solder various bits of other aluminium though and I'm quite satisfied with the results after getting to grips with the soldering process.
Warped due to soldering:
Time for a fittingLets see how big this thing gets. I laid out the septum boards on top of the aluminium sheets I have cut so far:
End piecesI decided to start on the end pieces as these have the most complex shape. After cutting the sheets and L profiles to length, I started drilling holes. I simply clamped the pieces together and drilled the holes. This way I have holes that line up perfectly without being to fussy about drilling precise. The key to this method is to number every part so it can be mated with the parts where drilled at the same time.
Ready for drilling:
![](https://i.ibb.co/qjPyzhD/20221203-141713.jpg)
There was a bit of a screw-up though. The sides of a pyramid should have an equal length and this is not the case. It turned out that I only added 22mm to the height of the side plate at the wide end and not at the narrow end (tip) of the pyramid. That is problem one. The second problem is that originally I wanted to bolt the sides of the pyramid together as well. I kind of assumed the angle would be 90 degrees so I could use an L profile. The angle is different. After some Googling I found the following page with the formula to calculate the angles:
https://math.stackexchange.com/questions/2263882/angles-between-lateral-faces-of-any-rectangular-based-pyramidF*** up detected:
![](https://i.ibb.co/XD9PDDG/20221127-160911.jpg)
The spreadsheet also calculates these angles but for my TEM cell it is too late.
The end pieces will also need cover plates with a hole through which the SMA connector on the septum sticks out. That hole needs to be at an angle because due to the offset of the septum, the tapered piece will be at an angle coming towards the cover plate. I decided to make small 'pans' with edges to get more mechanical strength. In order to transfer the holes on the pyramide I used the old pencil over paper trick to get a print of the holes. After that I simply aligned that so it would be straight on the pyramid. As a cherry on top, I soldered the edges of the pans shut so they won't fold that easely.
![](https://i.ibb.co/6RzhZhK/20221204-215209.jpg)
![](https://i.ibb.co/VJgg5XF/20221210-145237.jpg)
![](https://i.ibb.co/F0Kx0D5/20221210-150331.jpg)
![](https://i.ibb.co/XSgLkKF/20221210-161652.jpg)
Since the septum has an offset, it will arrive at the ends at an angle. The aluminium sheet is soft as snot so I bolted a rod into the hole for the SMA connector and just pushed it over to create the right angle.
![](https://i.ibb.co/r2cJCkP/20230114-224427.jpg)
Two end pieces done:
![](https://i.ibb.co/7W7Lzcm/20221210-143151.jpg)
Or not yet... The end pieces turned out to be rather flexible. Since the strength of the TEM cell comes from the end pieces, I decided to add braces to the side made from 20x10x2mm rectangular tube. This improved things a lot. But the root of the problem is that the sides of the pyramid aren't bolted together.
Door side panelAs I wrote before, I designed for a door that is 80 x 21 cm in size. I decided to make the door and side panel from one sheet. So the piece of sheet that is cut out from the side panel will become the door. I used my new plasma cutter that. What I forgot is that putting heat into sheet aluminium warps it. Let's hope that it will become flat again. Now the door needs to be RF tight so I designed a frame around the door that is cm2 in height. The door itself also has a frame around it that is 15mm in height. In addition the door also gets a handle (this is a bit of profile I had leftover). Between the door and the outer frame is a space of 3mm for contact fingers (from Wurth). I used 2.5mm rivets and double sided tape to fixate the EMC fingers. The fingers still make good contact despite the double side tape. Actually, some brands offer self-adhesive EMC finger strips. I used my new aluminium soldering skills to make the door frame and various door pieces. I used a so called piano hinge to hinge the door. This pretty RF tight and easy to install. In hindsight I wish I had used thicker L profile for the outer frame; I have this on hand but didn't think to use it. The pressure from the contact fingers does warp the frame a little bit.
![](https://i.ibb.co/NxCdRtT/20221217-173239.jpg)
![](https://i.ibb.co/NWXtJHR/20221217-173244.jpg)
![](https://i.ibb.co/LCT7Nj7/20221218-183328.jpg)
Connector panelThe connector panel is basically an extra plate on top of the side panel to make the side panel stiffer. It is held in place by 2.5mm rivets. I also cut it with the plasma cutter. Unfortunately the cut wasn't perfectly straight. Never mind.
![](https://i.ibb.co/hHyGkQn/20221211-163109.jpg)
Mid sectionThe mid section is the easiest part. I added a few ribs to the top, bottom and rear sheets and rivetted the L profiles (that form the flanges) to the sheet. Easy peasy. I did go through small packs of rivets like grazy though and the local shops where all sold out. I ordered a pack with 1000 rivets from Amazon. This was the cheapest option given I needed at least a couple of hundred more.
Time for putting everything together but lets clean all the parts first.
![](https://i.ibb.co/n8Dxsp4/20230114-201621.jpg)
Getting there...
SeptumThe septum is suspended from the top. There is 11cm between the top and the septum. I bought threaded plastic standoff from Aliexpress. Ordering these from Farnell / RS is insanely expensive. I connected two 40mm and one 50mm standoffs together to get to 11cm. Both sides have internal threads. Short 5mm standoffs serve as screws so I can tighten these with socket wrench.
![](https://i.ibb.co/NrR6QVq/20230114-224007.jpg)
Connecting the septum boards.
![](https://i.ibb.co/LnPpzww/20230115-201921.jpg)
I got some inspiration for creating the septum from a report that suggested to create the septum as strips that run along the length of the TEM cell. Resonating HF currents circle inside the TEM cell. The strips however, don't have a low impedance path to allow this current to flow and thus the strips dampen the resonances. To make the strips look like one plane, there are 39 Ohm resistors placed between the strips.
Assembly done!![](https://i.ibb.co/xzC4RkK/20230115-211731.jpg)
I specifically designed the door so I can make measurements at the top of the septum (where the field is more homogenous) and below the septum (where there is more space).
![](https://i.ibb.co/J7xKXqT/Inside.jpg)
And a close-up of the connector panel:
3 Measurements & tweakingImpedanceFrom my initial Excel sheet (not the one linked to this post!) I expected the impedance to be 55 Ohms. However, the actual impedance (according to a TDR measurement using an oscilloscope) turns out to be around 41 Ohms. That is much lower than I originally calculated! Time to sit down and think. The 'Crawford' document has some graphs on page 40 showing impedances for various geometries with offset septums (the Excel sheet calculates the axis values for these graphs as well). The graphs seem to be bang on with the 41 Ohm I measured. So I must be doing something wrong... After digging a bit deeper I found out I made a wrong assumption namely that with an offset septum the impedance would be driven by the smallest distance between septum and top/bottom. That turns out to be wrong; that is problem one. The second problem is that the 'Crawford' document says the term deltaC/Er can be ignored but that is not the case according to reference 2.
For the TDR measurement I used the calibrator output of my Lecroy Wavepro 7300 and a low-Z probe to measure the signal. An attenuator makes sure to have a somewhat reasonable 50 Ohm match
![](https://i.ibb.co/0nxbfzR/20230201-001816.jpg)
![](https://i.ibb.co/V99KwQk/Le-Croy13-orig-septum.jpg)
Reference 2 shows this formula:
\$Z_{0} = \frac{ \eta_{0}}{4[ \frac{a}{b} - \frac {2} {\pi} \ln ( \sinh (\frac {\pi g } { 2b})] - \frac {\Delta C } {\varepsilon_ { 0}}}\$
The term
\$\frac {\Delta C } {\varepsilon_ { 0}}\$ is calculated seperately in the Excel sheet. See reference 2 for the formulas.
With the term deltaC/Er taken into account, the calculated impedance (for a symmetric septum) is 45 Ohms instead of 55. Closer to 41 already but that could be just luck. Unfortunately this document doesn't show an easy formula for an offset septum. Extrapolating between the graphs that show results for septum offset ratios of h/b=0.2 and h/b=0.4, the w/a ratio for my TEM cell needs to be around 0.73 in order to end up a little bit over 50 Ohm. The reason to aim for a few Ohms over 50 Ohm is because having a DUT inside the TEM cell is likely to lower it's impedance and thus get close to 50 Ohms for real measurements. I would need to make the septum about 36 cm wide instead of 42 cm to get the w/a ratio in the 0.73 ballpark. As a test I tried disconnecting two strips in the septum (resulting in a width around 38-ish cm) in an attempt to make the impedance higher. This resulted in an impedance of 45 Ohm. Unfortunately having conductive strips inside the TEM cell, adds several more resonance points. All in all it looks like a need a different septum.
![](https://i.ibb.co/crRgSHq/Le-Croy15-fixed-septum.jpg)
But lets try do some math first. I did some more research and a document called 'Characteristic Impedance of a Rectangular Coaxial Line with Offset Inner Conductor' should contain a better method to determine the impedance of my TEM cell design but unfortunately it sits behind an IEEE paywall.
https://ieeexplore.ieee.org/document/1129506. After spending 40 euro's I got a PDF containing 8 pages with a bunch of formulas to digest. I was hoping to get a more in-depth report. Not just because of the money but also to achieve my goal. The 'Characteristic Impedance of a Rectangular Coaxial Line with Offset Inner Conductor' paper shows a number of integrals that lead to the impedance of a TEM cell. It also shows how the formula from reference 2 is derived once more but no direct formula that deals with an offset septum.
I could fire up Matlab and try to plug the integrals in but I preferred to find a way so Excel is able to calculate impedances for 50 OHm TEM cells with (about) any reasonable geometry. So I decided to try and figure out a more numerical approach myself and see if the graphs can be converted into an offset factor somehow. When playing with the numbers a bit and looking at the graphs it dawned to me that the graphs with the various h/b factors are basically the same exponential graph with a shift. If I can somehow come up with a factor I can add into the existing formula that calculates the impedance to also allow for offset septums, I'd be a happy camper. The exponential part is easy to spot in the formula so after I bit of tweaking I came up with the following formula which seems to fit nicely with the graphs:
\$Z_{0} =\frac{\eta_{0}}{4[ \frac{a}{b}-\frac{2}{\pi }\ln(\sinh(\frac {\pi g(1-1.41\frac {h} {b}^2)} {2b})]-\frac {\Delta C} {\varepsilon_{0}}}\$
I added the term \$(1-1.41 \frac {h} {b}^2 )\$ to include the influence of the offset septum. Again, keep in mind these formulas are approximations to come up with a simple way to calculate TEM geometries. The factor 1.41 is good for 50 Ohm impedances. For different impedances, it needs to be changed (which can be done by using the graphs from reference 1 and some trial & error)
Using the new & improved formula
![Cool 8)](https://www.eevblog.com/forum/Smileys/default/xcool.gif.pagespeed.ic.cOxV6sPNRd.png)
, it turns out I need a septum that is 35.5cm wide.
Outside leakageThis is what got the project started: a shielded measurement environment:
![](https://i.ibb.co/r0H8Dxm/Test-017-leakage.jpg)
Yellow trace is with the door open, orange with the door closed and blue with all the slits taped shut using aluminium tape.
ResonancesI did some measurements and as expected there are several frequencies that show resonances.
VSWR:
![](https://i.ibb.co/CsbdwL8/Test-019-VSWR-5-M-900-M.jpg)
Dampening from one side to the other:
![](https://i.ibb.co/Dt9FtqL/Test-021-attenuation-340-M-1500-M.jpg)
Upon closer inspection it looks like these resonances are not extreme; the peaks shown in the screen dumps are the maximum levels.
4 Immunity testingI have not gotten to setting up immunity testing. For that I want to use my R&S SMIQ RF generator and an external amplifier. 4W should be more than enough to get to field strengths 100V/m. But I'll need to write remote control scripts to control the RF generator but also create calibration tables to compensate for the amplifier gain. I've seen some interesting 4W RF amplifiers that work up to 1GHz on Aliexpress which look fine for this purpose.
I would also need some way to verify the field. I have added some calculations to the Excel sheet that show the required power / dBm levels to reach certain field strengths but these need to be verified using a field meter. Also notice that an offset septum helps to make a TEM cell that has to option to create very high field strenghts at the top where the septum is close to the ceiling of the TEM cell.
5 UsageLast but not least the TEM cell is supposed to be useful. In the meantime I have used it for various EMC related tests and I have to say I'm pretty happy with it. The results are similar of tests done in a real chamber and there is very little interference from outside noise.
This is a measurement I made to hunt down the source of a spike around 930MHz. This is right where the LTE bands are located so measuring without a chamber would mean trying to find it between the LTE towers blasting at high power levels.
6 ImprovementsThe most imminent improvement I need to do is get the septum sorted. Another happy day for JLCPCB!