EMC precompliance testing setup

[tszaboo]

I have some budget to set up an EMC precompliance testing setup at work. We've been launching products at a rate that this makes sense rather than immediately sending it to the test lab. Which I've been saying only for years now.

I wanted to get your idea about what equipment to add to this, or what's useless.
While it would be nice to just get a R&S 26GHz analyzer, it's going to be out of budget. I'm also a lot more interested in emission testing than immunity testing. What am I testing for? There are some specific standards, and generic ones. I'm targeting IEC 61000, industrial environments as a baseline for everything.
So far I'm looking into these parts:

SA: I'd like to cover all LTE signal bands, and look at a harmonics for some of them. 5 GHz is probably a good target. I think -160 dBm/Hz is a minimal sensitivity for such measurements. This is probably the highest cost of all the things, because I cannot get an anechoic chamber into the office. And I likely have to test more than one devices to see which one works.

Siglent SVA1075X, which has extra VNA functionality
Rigol RSA5065 with EMC as default now. This one ahs RTSA and VNA both of this is useful otherwise in the lab
Uni-t UTS3060A (very little experience with these as of now)
R&S FPL1007, maybe FPL-EMI7
Used equipment? There are too many to list. It should be supported and calibrated that limits the list.

And Tekbox EMCView software. I'm not sure I understand if the EMCView is a replacement for the EMC software on the one on the SA. The SA should be equipped with EMI options.
Antenna:
For me it looks like a Log periodic antenna is needed for low frequencies, up to about 2GHz, and horn antenna for higher frequencies. I'm not sure at the moment what's the benefit of biconical antennas.

TEM cell, to test the DUT with less environmental noise
LISN for DC voltages/currents
Near field probe set with Preamp for it
I'm not exactly sure what Comb generators are used for. As I understand this is to calibrate the antenna setup quickly? Can the tracking generator of the SA be used for this?
Cables to connect it all
Lab grade Adapter kits from N to SMA, BNC
A 20dB attenuator and a DC block

An EMC tent would be nice, if it fits the budget. But maybe I'll end up with also buying a Ecoflow or some other battery pack to be able to drive to the middle of nowhere to do the tests.

What's not covered here are obviously ESD and surge immunity testing. I'm treating those topics separately, as it is out of scope for many devices that I make.

[tautech]

Is pushing to 5 GHz really necessary for EMC ?

FWIW a few comments on the Siglent analyzers.
Preamp is a standard feature in all models.
EMC Mode is available in SSA3000X Plus, SSA3000X-R and SVA1000X ranges but it is an option that times out after ~120hrs of analyzer use.
Near field probes are available: https://www.siglenteu.com/accessory/srf5030t/

Attached a few screenshots from my SVA1032X....unfortunately they didn't capture the mouse pointer I placed to show the next menu to be clicked.

[tszaboo]

Is pushing to 5 GHz really necessary for EMC ?

FWIW a few comments on the Siglent analyzers.
Preamp is a standard feature in all models.
EMC Mode is available in SSA3000X Plus, SSA3000X-R and SVA1000X ranges but it is an option that times out after ~120hrs of analyzer use.
Near field probes are available: https://www.siglenteu.com/accessory/srf5030t/

Attached a few screenshots from my SVA1032X....unfortunately they didn't capture the mouse pointer I placed to show the next menu to be clicked.

5Ghz is, because I have radio devices that operate at these frequencies. LTE or 2.45GHz ISM. And not only radio modules.
If I would do complete due diligence, I would need to look at maybe the first 5 harmonics on my 2.45GHz transceiver.

Good point about the EMC mode timeout. I think I can do all the tests and worry about buying the license later. Though Rigol offers these as default now. After reading a lot about it, an RTSA might be lot better for EMC, if it supports these two at the same time. I'm trying to understand exactly how much, but from what I understand it reduces the testing time by an order of magnitude? Which can take hours.

[Hawaka]

To know what you need look at what exact standard you must conform. Each standard will tell you the exact required setup.

A few comments based on experience:
- For radiated emissions, you’ll need an anechoic chamber. You wouldn’t believe how much noise there is around. Technically you can just buy sheet metal and put it on the 6 side of a room and ground it. For a rough measurement it might be enough and it should be pretty cheap.
- Near-field probe are nice to find the source of noise, and that’s it. They won’t help to quantify the impact on the spectrum during testing.
- Definitively buy a calibrated noise source. That is the only way to confirm your entire measurement setup is correct.

- For conducted emissions, you’ll need a LISN. No need for the extra clean source. If your device pass test with the standard grid, then I’ll will probably pass with a cleaner source.

- I would definitively buy an ESD gun. It’s very easy to do the test in-house and also the most easy way to fail at the test lab.
- I wouldn’t worry to much for radiated and conducted immunity. If your design is properly grounded, and you have some properly dimension protection (like surge protector, TVS diode, ..) it should be fine.

- Plan from the start for automation, unless you only do a few measurement per year.

- EMC software are far from easy to use and license are expensive. Make to sure try a few and that it works with your spectrum analyser / setup. See what you like before committing. You can do without special EMC software, but that’s a hassle. You’ll have to look in the standard for every setup, every limit and so on. Then you are never sure you did it all correct. I would plan for it in the budget.

[tszaboo]

To know what you need look at what exact standard you must conform. Each standard will tell you the exact required setup.

A few comments based on experience:
- For radiated emissions, you’ll need an anechoic chamber. You wouldn’t believe how much noise there is around. Technically you can just buy sheet metal and put it on the 6 side of a room and ground it. For a rough measurement it might be enough and it should be pretty cheap.
- Near-field probe are nice to find the source of noise, and that’s it. They won’t help to quantify the impact on the spectrum during testing.
- Definitively buy a calibrated noise source. That is the only way to confirm your entire measurement setup is correct.

- For conducted emissions, you’ll need a LISN. No need for the extra clean source. If your device pass test with the standard grid, then I’ll will probably pass with a cleaner source.

- I would definitively buy an ESD gun. It’s very easy to do the test in-house and also the most easy way to fail at the test lab.
- I wouldn’t worry to much for radiated and conducted immunity. If your design is properly grounded, and you have some properly dimension protection (like surge protector, TVS diode, ..) it should be fine.

- Plan from the start for automation, unless you only do a few measurement per year.

- EMC software are far from easy to use and license are expensive. Make to sure try a few and that it works with your spectrum analyser / setup. See what you like before committing. You can do without special EMC software, but that’s a hassle. You’ll have to look in the standard for every setup, every limit and so on. Then you are never sure you did it all correct. I would plan for it in the budget.

I know how much noise is around, since I took some portable Anritsu around with some antenna. I saw for example -60dBm noise in the 8-900MHz region. Directional antenna helps a lot.

LISN: DC only, and even that's rare. I have nothing with AC source, it's not very compatible with the ATEX directive.

ESD I'm not worried at the moment. The devices that I'll be sending out for testing have interfaces that are tested already on similar devices. In fact I don't remember ever failing surge and ESD. I failed immunity few years ago on a device that had Ethernet port. It failed 100MHz immunity. Like, no surprises there, right, in band large noise signal?

I wouldn't even start without software. I can only imagine that report generation for CE would take days to do. I have better things to do with my time than that.

[tszaboo]

- Definitively buy a calibrated noise source. That is the only way to confirm your entire measurement setup is correct.

I was thinking something like this:
https://www.tekbox.com/product/tbcg1-radiating-comb-generator

Seems cheap enough to include. Especially since I'm looking at paying ~2000 EUR just for antennas. And then I'm still buying cheap antennas.

[Hawaka]

I have never heard of tekbox, so I cannot comment.

I’ve used Teseq or Schwarzbeck accessories, pricey but they do the job.

Measurement confidence is really important for EMC, IMO you can cheap out on the full measurement chain, and buy a good source to characterize your setup. Or you buy a good setup and you still get a good source to test it.

PS:
Schwarzbeck has a used equipment section on their website, might worth a look

[2N3055]

I have Tekbox near field probe set with preamp.
Works well and had no problem for years.

I remember someone testing that Comb gen and they said it was OK. Certainly worth the money.

[tszaboo]

I have never heard of tekbox, so I cannot comment.

I’ve used Teseq or Schwarzbeck accessories, pricey but they do the job.

Measurement confidence is really important for EMC, IMO you can cheap out on the full measurement chain, and buy a good source to characterize your setup. Or you buy a good setup and you still get a good source to test it.

PS:
Schwarzbeck has a used equipment section on their website, might worth a look

I didn't know about Schwarzbeck. I've checked the price of some of their antennas, it's a lot. I get it, quality.
Teseq I know, I think I've seen that being used for compliance testing. Again, budget issues.
Tekbox seems to have a good range of devices, and software that seems to geared towards Rigol, Siglent and low end RS SAs. I wouldn't be surprised to just buy everything from them.

Right now I'm looking also at Aaronia. Some of their log-antilog antennas are only a few hundred EUR, very reasonably priced. They also seem to have some USB RTSA, that's well priced, except the software. Like the base device with 1RX / 1TX is 4200 EUR on their website, but the software option to use the TX as a tracking generator is 2500 EUR  :'(

[nctnico]

- EMC software are far from easy to use and license are expensive. Make to sure try a few and that it works with your spectrum analyser / setup. See what you like before committing. You can do without special EMC software, but that’s a hassle. You’ll have to look in the standard for every setup, every limit and so on. Then you are never sure you did it all correct. I would plan for it in the budget.

I have solved this with some Python scripting which controls the instruments and produces plots with the limit lines in them. Based on real EMC testing results I have devised some correction values for the TEM cell and antennas I'm using. IMHO having EMC options is overrated. A test setup somewhere in a lab has a lot of sources for error to begin with. Having the bandwidth at about the same setting as the official standard is good enough. Apply averaging of a measurement to see where the pain points (frequence peaks exceeding the limit) are. If a peak stays there with averaging, then it will remain there with a quasy peak measurement as well. Don't expect to get to within a few dB of the real / actual test. The anechoic chamber and antennas used in the real test will result in different measurement results.

When testing radio devices, a 5GHz spectrum analyser isn't going to cut it. CE requires testing to around 13GHz (IIRC) and FCC to around 21GHz. 2.4GHz radio chips can produce a surprisingly large amount of harmonics at 7.2GHz.

I also have a comb generator. Unfortunately it starts at 100MHz. Big antennas are typically pretty flat from 100MHz already. If I'd buy another one, I'd get one which starts at 10MHz because you'll want to measure radiated emissions from 30MHz and you'll need to have some reference at that frequency to determine the antenna response if you need to do an absolute measurement.

[EEVblog]

The low end Rigol SA with EMC software option and a 30MHz to 1GHz biconical antenna has been a popular low cost in-house pre-compliance setup for a long time now.
The usual radiated emmision standards are that 30MHz to 1GHz range, but if you have something specific beyond that you might need more specialist advice. I think 6GHz is the next step-up in range.
TekBox are a popular maker of low cost EMC testing gear, hard to go wrong there.

[tszaboo]

- EMC software are far from easy to use and license are expensive. Make to sure try a few and that it works with your spectrum analyser / setup. See what you like before committing. You can do without special EMC software, but that’s a hassle. You’ll have to look in the standard for every setup, every limit and so on. Then you are never sure you did it all correct. I would plan for it in the budget.

I have solved this with some Python scripting which controls the instruments and produces plots with the limit lines in them.
...
IMHO having EMC options is overrated.
...

While I appreciate the effort to make this sort of testing as cheap as possible, I think I'll take the "Nobody got fired for buying HP" option. A full scan could take hours from what I see, and I just don't think I can afford spending days chasing bugs in a python script. Not to mention the standards. I see on a previous product I have, they used TS151.010-1 GSM as emissions limit. So I just downloaded the standard, it's 6127 pages long.

FCC defined with their infinite wisdom, the upper limit of radio testing as 3000 GHz.
You are right about the spurs in 2.4GHz radios. We had a 20 something GHz SA rented and made conducted emissions testing on our radio interface. Basically desoldered the antenna, and wired the SA directly into it. Passing that actually gives me quite a bit of confidence.

The low end Rigol SA with EMC software option and a 30MHz to 1GHz biconical antenna has been a popular low cost in-house pre-compliance setup for a long time now.
The usual radiated emmision standards are that 30MHz to 1GHz range, but if you have something specific beyond that you might need more specialist advice. I think 6GHz is the next step-up in range.
TekBox are a popular maker of low cost EMC testing gear, hard to go wrong there.

Biconical seems to be the equipment of choice for low frequency measurements.
While I am not trying to make my setup as cheap as possible, once I added everything together it's about 6-7000 EUR just for the accessories, and about the same for the SAs I listed. Which seems like a lot, but it's less than a FCC+RED certificate.

- For radiated emissions, you’ll need an anechoic chamber.

For the "anechoic chamber", probably the best I can do it to use an intermodal container. Did anyone try to make a test in one of those? In my mind now, I'll get reflections, but reduced environment noise.

[tszaboo]

Grok 4

Can a intermodal container be used as a makeshift anechoic chamber for RF measurements? How good isolation can we expect? How bad are the reflections?

Yes, an intermodal (shipping) container can be repurposed as a makeshift anechoic chamber for RF measurements, but it requires significant modifications to function effectively. Without alterations, a bare steel container acts more like a reverberation chamber due to its highly reflective metal walls, leading to severe internal reflections and standing waves that would render precise RF testing (e.g., antenna patterns, emissions, or immunity) unreliable. With proper shielding enhancements and internal lining using RF-absorbing materials, it can provide a semi-anechoic environment suitable for pre-compliance testing, especially at higher frequencies. Commercial examples exist where containers are custom-engineered into portable RF chambers for this purpose. Below, I'll break down the feasibility, expected isolation, and reflection performance based on available designs and discussions.
Modifications Needed

Shielding for Isolation: Seal seams, joints, and doors with conductive tape (e.g., copper or aluminum) to create a Faraday cage effect. Install RF-shielded doors, filtered power/communication penetrations, and ventilated panels with RF seals to prevent leaks. Without these, gaps in the container's construction (e.g., corrugated walls and non-RF-optimized doors) allow external signals to penetrate.
Absorbers for Reflection Control: Line the interior walls, ceiling, and floor with RF-absorbing materials like pyramidal foam (e.g., 12-18 inch microwave absorbers made from carbon-loaded polyurethane) or hybrid setups with ferrite tiles for lower frequencies. These convert incident RF energy into heat, reducing echoes. DIY alternatives include conductive foam, charcoal-iron powder mixtures in paint or mud, or even saltwater layers for basic absorption, though these are less effective and harder to optimize.
Other Considerations: The container's limited dimensions (e.g., internal width ~7'8", height ~7'10" for a standard 20-ft unit) restrict the "quiet zone" for testing, potentially causing near-field effects at lower frequencies. Placement in a low-RF-noise area (e.g., a field) helps, and costs for retrofitting can range from a few thousand dollars for basic DIY to tens of thousands for professional setups.

Expected Isolation (Shielding Effectiveness)
Isolation refers to how well the chamber blocks external RF signals, measured in decibels (dB) of attenuation. A modified shipping container can perform well here due to its steel construction:

Bare Container: Provides partial Faraday cage shielding (e.g., 50-60 dB attenuation in some frequency bands), but leaks at seams and doors reduce this, allowing ambient noise (e.g., FM radio, cell signals) to interfere. Not suitable for sensitive measurements without fixes.
With Modifications: Achieves 90-125 dB or more shielding effectiveness across a broad frequency range (e.g., 100 MHz to several GHz), sufficient to drop external signals below the noise floor of most test equipment. This is comparable to foil-lined rooms or commercial shielded enclosures. Performance is tested using methods like normalized site attenuation (NSA) or by measuring signal differences inside vs. outside at frequencies like 1 GHz. Filtered penetrations are critical—unaddressed ones can drop isolation by 20-40 dB.
Frequency Dependence: Better at higher frequencies (above 200 MHz); lower frequencies (below 100 MHz) may require additional ferrite layers or thicker shielding to maintain high isolation.
Limitations: Corrugated walls can create unpredictable leakage paths, and without professional installation, real-world isolation might be 70-100 dB. For comparison, full commercial anechoic chambers aim for 100+ dB.

In practice, this level of isolation makes a container viable for pre-compliance RF testing in noisy environments, but it's not as robust as a purpose-built chamber.
Expected Reflections (Absorption Performance)
Reflections are the bigger challenge in a makeshift setup, as metal walls cause RF waves to bounce repeatedly, creating resonances and measurement errors. Performance is quantified by reflectivity (dB attenuation of reflected waves) or NSA deviation (how closely the chamber mimics free space, ideally +/-4 dB for compliance).

Bare Container: Extremely poor—reflections are near 100% (0 dB attenuation), leading to high Q-factor resonances (e.g., fundamentals at 30-60 MHz in a typical container) and standing waves that can distort measurements by 20-30 dB or more. This turns it into a reverberation chamber, useful for some immunity tests but useless for precise anechoic work.
With Absorbers: Significantly improved, but frequency-dependent:

Above 200 MHz: Pyramidal foam absorbers (12-18 inches) can achieve 10-20 dB reflection attenuation, leading to NSA deviations of +/-6 dB. This is adequate for antenna testing or emissions measurements in the GHz range (e.g., Wi-Fi, 5G).
Below 200 MHz: Without ferrite tiles, reflections worsen to +/-20 dB NSA deviation due to poor absorption. Adding ferrite (e.g., 3-10 mm tiles) can improve this to +/-4-6 dB but adds weight (~10 lb/sq ft) and cost.
DIY Options: Materials like conductive foam or saltwater (e.g., 3mm layer of 26% NaCl solution) might provide 10 dB attenuation at 100 MHz-1 GHz, but results are inconsistent and require testing. Charcoal-iron mixes offer marginal improvements (e.g., 0.1-5 dB at low frequencies).


Overall Quality: Expect reflections to be "bad" (10-20 dB worse than ideal) without full coverage absorbers, limiting accuracy for far-field measurements. For small-scale tests, a quiet zone of ~30x30x30 cm might be achievable with basic lining, but larger containers need extensive material to cover resonances from wall corrugations.

In summary, a modified intermodal container can serve as a cost-effective makeshift chamber with good isolation (90+ dB) and manageable reflections (10-20 dB attenuation above 200 MHz), but it's best for pre-compliance or higher-frequency work. For critical applications, professional chambers are preferable due to better consistency. If pursuing this, start with sealing for isolation, then add absorbers iteratively while testing with a network analyzer.

So the container without RF absorber inside probably would act as an echo chamber, not an anechoic chamber. I somewhat expected that. I definitely cannot add ferrite tiles and cover every hole on it. Probably what I could do is test the noise in the container, compare it with outside to see the isolation.
Next step, place the Comb generator in place of the DUT, and compare it with open field results, and with the comb generators calibration sheet, to see how bad reflections are. Without lining the entire container with foam absorbers, I'm guessing immediately behind the device looking at it with the antenna, and the parallel sections should be covered with absorber.

[nctnico]

- EMC software are far from easy to use and license are expensive. Make to sure try a few and that it works with your spectrum analyser / setup. See what you like before committing. You can do without special EMC software, but that’s a hassle. You’ll have to look in the standard for every setup, every limit and so on. Then you are never sure you did it all correct. I would plan for it in the budget.

I have solved this with some Python scripting which controls the instruments and produces plots with the limit lines in them.
...
IMHO having EMC options is overrated.
...

While I appreciate the effort to make this sort of testing as cheap as possible, I think I'll take the "Nobody got fired for buying HP" option. A full scan could take hours from what I see, and I just don't think I can afford spending days chasing bugs in a python script.

I get that but having your own suite of tools gives a lot of flexibility. There will be a learning curve to use bought software as well with plenty of room for error. It also depends on what your goal is. I like to start doing checks during the design phase as part of testing prototypes. But I'm not going to wait hours to do a single measurement. Especially when trying components to see which one gives the best performance from an EMC perspective versus price. I just fire up the spectrum analyser, flip the prototype into the TEM cell and run a couple of sweeps. Sometimes over a limited frequency band. The (command line) script I've made supports doing quick measurements like that as well as doing more elaborate measurements. I only do a longer test once I'm sure the prototype behaves as it should. But still I'm not going to do extremely long measurement runs from every height / angle like they do during a real test. This may sound like not covering all possible problems but in my view it doesn't make sense to take a product to a test lab which barely passes pre-compliance testing at my own lab. If a product stays well under the limit when measured sitting normally, rotated over the X and Y axis in a TEM cell, there is no reason to fear a pre-compliance scan at a test lab.

If you are going to buy software, I'd look at how easy it is to define measurements, influence the measurement speed (like by setting the number of averages, changing RBW), recall existing measurements and whether you can define your own tests (for example to do RED testing).

Not to mention the standards. I see on a previous product I have, they used TS151.010-1 GSM as emissions limit. So I just downloaded the standard, it's 6127 pages long.

Where it comes to standards my approach is to either find a document or somebody to tell me which test to perform -exactly- and what the limits are. Then again, I'm not pursueing a career to know everything there is about EMC related standards.

Either way, part of this journey will be about building your own confidence in your measurement results.

[kmoonwalker]

if you are going to be at 5 GHz region I think that one needs to consider some new or preowned rohde as from personal experience some of companies asking you to manufacture or test something for them will specifically demand it to be Rohde or Keysight in some cases and I think it is good to seach for various options but it is also good to reach out to them and ask how they look at this cause I seen multiple times that customer after just seeing the logo or equipment name on test report called in to have the test equipment replaced due to being non conformant with their internal policy... 

[tszaboo]

if you are going to be at 5 GHz region I think that one needs to consider some new or preowned rohde as from personal experience some of companies asking you to manufacture or test something for them will specifically demand it to be Rohde or Keysight in some cases and I think it is good to seach for various options but it is also good to reach out to them and ask how they look at this cause I seen multiple times that customer after just seeing the logo or equipment name on test report called in to have the test equipment replaced due to being non conformant with their internal policy... 

Well, we'll have CE declaration in the end. Technically they rarely ask for test report, and even if they do it's none of their business how it was tested.
If you would be a consultancy, where your end product is the design, and you have one customer it's a different story.
But yes, there are some picky customers even in B2B, who want to have ie factory visits before buying stuff.

[tszaboo]

I'm adding the following tools into the list:
GND plane metal plane for LISN
Set of ferrite cores, Würth 742711
N connector transient limiter

With a Power amplifier and the tracking generator of the SA, and TEM cell, it's apparently possible to do some immunity testing. This would only need the power amplifier as the extra part.
Well, and a coupler to monitor the RF output.

[jusaca]

After reading a lot about it, an RTSA might be lot better for EMC, if it supports these two at the same time. I'm trying to understand exactly how much, but from what I understand it reduces the testing time by an order of magnitude? Which can take hours.

But does such an RTSA offer the small RBW needed when using large spans?
I think a common EMC standard RBW is 9 kHz. The Siglent / Rigol RTSAs can capture 40 MHz in one piece, but have only around ~1000 bin. At least thats how I understand their documents.
And 40 MHz with 1k bins means a spacing of the measurement points around 40 kHz apart. Not sure if this is sufficient.

[tszaboo]

After reading a lot about it, an RTSA might be lot better for EMC, if it supports these two at the same time. I'm trying to understand exactly how much, but from what I understand it reduces the testing time by an order of magnitude? Which can take hours.

But does such an RTSA offer the small RBW needed when using large spans?
I think a common EMC standard RBW is 9 kHz. The Siglent / Rigol RTSAs can capture 40 MHz in one piece, but have only around ~1000 bin. At least thats how I understand their documents.
And 40 MHz with 1k bins means a spacing of the measurement points around 40 kHz apart. Not sure if this is sufficient.

I never used an RTSA. From what I see from Siglent's and Rigol's datasheet, their RTSA can operate in regular sweeped SA mode. And they also have, with the right option, the Quasy-peak filter that is required for the EMI tests.
If you look at the block diagram of an RTSA and a regular SA, they only deviate quite late in the signal path. I'm guessing they might be doing some of the traditional analog processing digital. Both these companies have filters, RBW down to 1Hz.
Here is some Tek appnote about this:
https://www.tek.com/en/documents/application-note/real-time-spectrum-analysis-emi-diagnostics

[jusaca]

Yeah sure, you can go down to really small RBW, I don't doubt the RTSA version can do the same and more than the non-RT version.
But can you combine the fast real time update over wide bandwidths (like 40 MHz) with small RBW? That is the point I doubt, so I'm not sure if it will really offer faster measurement speed for EMC tests.
At least I don't understand how. If they do really offer such vast improvements in speed with the needed resolution for standard-conform tests I would be interested in such a device as well

[scottapotamas]

I don't have the pro version of EMCview to use the RTSA bits of the Rigol, but just be aware it's not a clear win over swept mode.

The sensitivity isn't the same and there's some spurs and noise floor artifacts that aren't in the marketing material.

The integrated limit line and scan table libraries in the RSA's EMI app are fine, able to run sweep sequences with varing points/RBW across specified ranges pretty well. Loading the LISN's correction table isn't particularly hard either. I started that way but bought a EMCview licence for more consistent reporting/documentation.

Quasi-peak is slow as hell, so normally just run peak and average scans (good to keep some margin) and use QP for the largest peaks if needed.
 
You mentioned DC LISN over AC earlier in the thread - be aware of which inductance you're getting with the specific tests you'll be working against and the need to distinguish between CM/DM. The typical 5uH are useful enough for precomp but had larger variance from the test-house's results than I expected the first few times around, a 50uH may have been a better choice for me (or a mains one).

[tszaboo]

I don't have the pro version of EMCview to use the RTSA bits of the Rigol, but just be aware it's not a clear win over swept mode.

The sensitivity isn't the same and there's some spurs and noise floor artifacts that aren't in the marketing material.

The integrated limit line and scan table libraries in the RSA's EMI app are fine, able to run sweep sequences with varing points/RBW across specified ranges pretty well. Loading the LISN's correction table isn't particularly hard either. I started that way but bought a EMCview licence for more consistent reporting/documentation.

Quasi-peak is slow as hell, so normally just run peak and average scans (good to keep some margin) and use QP for the largest peaks if needed.
 
You mentioned DC LISN over AC earlier in the thread - be aware of which inductance you're getting with the specific tests you'll be working against and the need to distinguish between CM/DM. The typical 5uH are useful enough for precomp but had larger variance from the test-house's results than I expected the first few times around, a 50uH may have been a better choice for me (or a mains one).

Thanks, good to see that others have similar setups. I'm leaning towards the RSA5000N because of the EMC is included by default. On the other hand I can see the 3rd harmonic of the 2.4Ghz on the Siglent's 7.5GHz bandwidth. Decisions, decisions...
I think I'm fine with some spurs in precompliance testing. If it goes above the line, I have to investigate it anyway. I'm reading that some of the QP measurement, the low frequency ones, could take 4 hours to perform. That's a problem because that's a whole day gone for one test, I might need to do it in two orientation of the DUT and Antenna... I would rather spend my time on better things.

I'm trying the EMCview in demo mode. They have separate limit lines for average responding and peak filters? That sounds very useful.

[tszaboo]

Yeah sure, you can go down to really small RBW, I don't doubt the RTSA version can do the same and more than the non-RT version.
But can you combine the fast real time update over wide bandwidths (like 40 MHz) with small RBW? That is the point I doubt, so I'm not sure if it will really offer faster measurement speed for EMC tests.
At least I don't understand how. If they do really offer such vast improvements in speed with the needed resolution for standard-conform tests I would be interested in such a device as well

Testing houses nowadays have a dedicated EMI receiver.
From what I understand, those are faster by an order of magnitude. And have preselector filters for example, or you can set up your devices RF transmission time, so it samples when you transmit. If you don't have continuous wave available for a radio. It's also possible that these EMI receivers are nowadays built on the same principles, but then, it's super expensive equipment anyway.

[nctnico]

After reading a lot about it, an RTSA might be lot better for EMC, if it supports these two at the same time. I'm trying to understand exactly how much, but from what I understand it reduces the testing time by an order of magnitude? Which can take hours.

But does such an RTSA offer the small RBW needed when using large spans?
I think a common EMC standard RBW is 9 kHz. The Siglent / Rigol RTSAs can capture 40 MHz in one piece, but have only around ~1000 bin. At least thats how I understand their documents.
And 40 MHz with 1k bins means a spacing of the measurement points around 40 kHz apart. Not sure if this is sufficient.

IIRC the spacing between the steps (bins) may be halve the RBW at most. This is also why EMC measurements typically consist of multiple sweeps stitched together in order to satisfy the minimum frequency step size.

[nctnico]

With a Power amplifier and the tracking generator of the SA, and TEM cell, it's apparently possible to do some immunity testing. This would only need the power amplifier as the extra part.
Well, and a coupler to monitor the RF output.

For immunity testing you'll also want AM modulation. An electric field probe to verify the electric field strength is also nice to have but in a TEM cell you can calculate the resulting field strength from the input power and geometry. Don't forget a load capable of dissipating the amplifiers power. I opted to buy a 5W 20dB attenuator (a good quality one!); the attenuator can also be used to measure the transfer characteristic of the amplifier itself.

In my setup I measured the amplifier's frequency response at several power levels and created a calibration table. The table is used to control the power level of an RF generator (in my case an R&S SMIQ) to create the desired field strength through my script. There are also dwell time requirements for immunity testing (DUT response time to events). All in all I'm not sure a tracking generator on a spectrum analyser will do much good. You'll need software to run immunity tests because they are too tedious to do manually. From all the EMC tests, the immunity tests take most of the time. Count on at least 20 minutes per run.

Personally I like to test products up to 30V/m. For consumer stuff the limit is 3V/m but that is too low for flawless operation in the real world in my experience.