Parametric TEM cell design

[TopQuark]

I wanted an open TEM cell, but didn't want to pay the few hundred USD TekBox charges, so I designed my own. The CAD model is designed in Fusion 360 and is parametric, meaning you can punch in the length and width of the TEM cell you want, and the model will adjust itself to generate a functional TEM cell, that's the idea anyways.

The upper and lower two identical halves of the TEM cell are designed with the sheet metal process in Fusion 360, so you could create a drawing of the base shape of those parts and send it to your sheet metal fabricator of choice. Still debating if the septum (center sheet) should be made of copper sheet or PCB material.

The formula used to calculate cell height based on septum width and septum thickness is derived from http://eagle.chaosproject.com/sandbox/acstrial/newsletters/summer08/pp2.pdf . As a sanity check, I used dimensions from a couple of commercial TEM cells and ran it though the formula and the results are close to expected. If someone smarter than me can run the generated designs through EM simulations, that would be much appreciated.

Link to F360 Model: https://a360.co/3BWSWfj

UPDATE: Don't build this, the design is faulty and doesn't work properly. Read my update here: https://www.eevblog.com/forum/projects/parametric-tem-cell-design/msg5457338/#msg5457338

[TopQuark]

Production files for a TEM cell with 200x200mm rectangular septum area.

The cell top and bottom plates are made with 1mm 304 stainless steel plates, cut and bent to spec by a Chinese metal shop. DXF for cutting and PDF drawing for bending provided. You'll need 2 of these.
Septum is 1.2mm FR4 from JLCPCB. Gerbers provided.
Connectors are panel mounted SMA, I wanted to standardise my lab to use SMA for everything.

Everything together with shipping costed me less than 100USD, your mileage may vary.

[krokodyl1220]

Thanks for the design.
I am currently working on making the cell myself, but either I am calculating something wrong or the formula from the paper is incorrect. I was testing it on tekbox cells and I got about 50 ohms only after changing er to 2.2

I would like to see how you did the formula, but it seems like there is some error on Autodesk's side (can't open design in fusion, waiting for download link for 4 days...). Could you please export the model as fusion file and upload it here?

[TopQuark]

Apologies for the delay in response, I didn't check the thread until I wanted to update it today. I have attached the f3d file with this message.

I followed the formula from this page http://eagle.chaosproject.com/sandbox/acstrial/newsletters/summer08/pp2.pdf . After simplifying the formula, I arrived at the following result: Distance between septum and cell wall = SeptumWidth / 2.56826 + SeptumThickness / 2.

[TopQuark]

So I received the stainless steel parts today and built the TEM cell.

Measured VSWR with my librevna and it was terrible, I realised I have goofed up the septum design by cutting slots into it. I originally did this as I read from a paper this could extend the usable frequency range of the TEM cell, but perhaps my execution of the idea is flawed. I fixed my mistake by covering both sides of the septum with copper tape, essentially mending the slots I have cut into it.

After fixing the septum, I am getting VSWR < 2 under 2 GHz, which is good enough for me.

[krokodyl1220]

Great, thanks for the design file, I will see what I messed up

Also a tip for getting less noise - you can create a "closed cell" with just cardboard cut to fit on the tem sides, and aluminium foil glued to the outside and connected to tem enclosure. The measured emissions from the device are similar to what you get without the shielded walls, but the difference in noise from the environment is massive.

As for the issue with slots - tekbox cells (at least TBTC3 that we have at work) also use splits in the septum lengthwise, but the sections are connected to each other using SMD capacitors.

[TopQuark]

Hi future and past visitors:

The formula used to calculate cell height based on septum width and septum thickness is derived from http://eagle.chaosproject.com/sandbox/acstrial/newsletters/summer08/pp2.pdf

So I received the stainless steel parts today and built the TEM cell.

Measured VSWR with my librevna and it was terrible, I realised I have goofed up the septum design by cutting slots into it. I originally did this as I read from a paper this could extend the usable frequency range of the TEM cell, but perhaps my execution of the idea is flawed. I fixed my mistake by covering both sides of the septum with copper tape, essentially mending the slots I have cut into it.

After fixing the septum, I am getting VSWR < 2 under 2 GHz, which is good enough for me.

Reflecting on the sloppy results obtained, careful re-examination of the source material referenced and further study of the subject matter, I realized my proposed design was faulty from an electromagnetic point of view and would not provide satisfactory performance. This is in addition to the problematic mechanical design of my TEM cell which created durability issues, further limiting the usefulness of my TEM cell design.

I regret the errors, and apologize for any wasted money and effort spent on replicating my unsuccessful design. Please don't build my TEM cell as proposed above, and use this thread as a study on how NOT to design a TEM cell.

I'll bring my new conversations here: https://www.eevblog.com/forum/projects/tekbox-tem-cell-how-is-it-constructed/

- Alex

[Miek]

For what it's worth, I built a TEM cell based off this design last year and had great results with it, so thank you! We did a few rounds of half-days at the EMC lab, and each time I could correlate the results pretty well to what I was getting in the TEM cell. It was much more repeatable than the usual near-field probes, so I could try out a bunch of different PCB/enclosure modifications and re-test with good confidence we'd see the same relative change in results at the next lab session.

I did make a few modifications to the design when I built mine: I changed the ends similar to the TekBox design so they would clamp the septum rather than rely on the solder joint to support it, and I didn't do any cuts in the septum. I made the septum out of single sided copper-clad, cut to shape on a bandsaw, and then sanded away the extra copper at the ends. I couldn't get a large enough single sheet so I just used copper tape (with conductive adhesive) to stick them together.

There was a mistake in mine where I tapered the septum copper down to a point but the shield doesn't taper right down, so there is some impedance mismatch at the ends (you can see the inductive bumps in the TDR plot below), but it worked so well for emissions testing I never bothered re-making it.

[Miek]

(extra photos that went missing from the previous post)

[tomud]

I regret the errors, and apologize for any wasted money and effort spent on replicating my unsuccessful design. Please don't build my TEM cell as proposed above, and use this thread as a study on how NOT to design a TEM cell.

I'll bring my new conversations here: https://www.eevblog.com/forum/projects/tekbox-tem-cell-how-is-it-constructed/

- Alex

There's nothing to regret. Personally, I was interested in the topic and decided to repeat the experiment.

When designing my septum, I modeled it on the TekBox septum (based on photos I found on the Internet - although there are not many of them).
I am currently waiting for the PCB I ordered. Once I get the PCB, I will try to reproduce similar effects (time will tell if I succeed).
In the meantime, I am looking for any scientific studies that will explain the theoretical basis of the solution used by TekBox.

Why do I want to follow the TekBox solution as an example? For a simple reason - looking at this chart, you can see that they achieved very good results.

https://essentialscrap.com/tem_cell/tem_cell_vswr.png