Author Topic: Best spectrum Analyzer ($ vs. perf.) for pre-compliance EMI test in home lab  (Read 6568 times)

0 Members and 1 Guest are viewing this topic.

Offline Jester

  • Frequent Contributor
  • **
  • Posts: 459
  • Country: ca
The title about sums it up "Best spectrum Analyzer (Cost vs. performance/features.) for pre-compliance EMI test in home lab."  Hope to keep this under $1k, might stretch to $2k if it makes sense.

Used is an option, how well suited is the Rigol SDA815?

What about Agilent E7495?

I already have some near field probes.
« Last Edit: July 25, 2016, 08:04:24 pm by Jester »
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 21236
  • Country: nl
    • NCT Developments
Did you look at the Advantest R3131?
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Wuerstchenhund

  • Super Contributor
  • ***
  • Posts: 3079
  • Country: gb
  • Able to drop by occasionally only
What about Agilent E7495?

My former E7495B is now used for EMC pre-compliance testing, and the new owner seems to be quite happy with it in that role.
« Last Edit: July 26, 2016, 07:01:44 am by Wuerstchenhund »
 

Offline Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 2541
  • Country: ca
Hi,

I would stay away from the Agilent E7495, they start at 10 MHz and you might need to measure conducted emissions  down to 150 kHz.

RBW

Analyzers used for EMC have RBW settings of 200 Hz, 9kHz and 120kHz. In addition to being different bandwidths, they have a different shape, than the normal 1-3-10 series filters. Not critical, but if you can get the right filters, why not.


QP

Some EMC analyzers have the ability to measure quasi peak. The quasi peak measurement simulates the measurement that would be made with an analog meter. This is because originally EMC specs were written for an EMC receiver.
The QP detector only help if the EMC is in short bursts, a relatively long time between bursts.
The rise time on the QP detector is 1ms, so if your emissions are around 1ms or shorter, you benefit from the QP detector.
The fall time is 160ms, so if the time between the emissions is 100ms or greater you will benefit.
The QP detector has no benefit if the emissions are constant.

QP is always less than the peak, and greater than the average.

I rarely do QP measurements.


see: https://www.eevblog.com/forum/testgear/rf-spectrum-analyzer-computing-quasi-peak/

I believe that Rigol SDA815, has an option for the EMC features.

I use a HP8595E (6.5 GHz) for my EMC measurements. The 2.9GHz HP8594E would be just as good. These do not have a QP detector, unless Option 103 is installed

LISN

You can build your own LISN. Have a look at this thread:

https://www.eevblog.com/forum/projects/5uh-lisn-for-spectrum-analyzer-emcemi-work/msg404662/#msg404662

This LISN contains a high pass filter, attenuator and limiter to protect the spectrum analyzer.

Regards,

Jay_Diddy_B

 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 17374
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
You can build your own LISN. Have a look at this thread:

https://www.eevblog.com/forum/projects/5uh-lisn-for-spectrum-analyzer-emcemi-work/msg404662/#msg404662

This LISN contains a high pass filter, attenuator and limiter to protect the spectrum analyzer.

Note if you're doing automotive, that's fine (you need two), but if you're doing mains, you need one more like this:



Which as you can tell, is no less home-buildable. :)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline skipjackrc4

  • Regular Contributor
  • *
  • Posts: 239
  • Country: us
Regarding the EMI-specific RBW bandwidths--it's not that critical to actually have them built-in to the instrument.  If you perform a scan with a lower RBW and a higher point density (sample spacing less than RBW/2) then you can basically convert to another RBW mathematically through deconvolution/convolution.  Going the other way (high RBW to low RBW) doesn't really work due to the increased noise floor.

In many cases, using a slightly "wrong" RBW won't matter at all; in others it matters quite a bit.  It depends on what the BW of the signal is in relation to the RBW.
 

Offline rf-loop

  • Super Contributor
  • ***
  • Posts: 3581
  • Country: cn
  • Born in Finland with DLL21 in hand
If practice and theory is not equal it tells that used application of theory is wrong or the theory itself is wrong.
-
Huawei HarmonyOS 2.0  |  ArcFox Alpha S
 

Offline Jester

  • Frequent Contributor
  • **
  • Posts: 459
  • Country: ca
Big thanks to all that responded, the comments about lower frequency capability for conducted emission testing, quasi-peak, RBW etc were all quite helpful. I'm inclined to order either the Rigol or Siglent with the EMI option.

I will make or purchase LISN. Is there any reason why a LISN designed for 120Vac would be unsuitable to also test battery or wall-wort powered devices?

 

Offline pascal_sweden

  • Super Contributor
  • ***
  • Posts: 1523
  • Country: no
Some more info about the EMI Measurement Kit:
http://www.emclive2016.com/2016/03/08/siglent-presents-ssa3000x-spectrum-analyzer-small-lightweight-user-friendly/#1462982494743-bb0f5249-2290

Has anyone on this forum actually bought and used this EMI Measurement Kit?
What are your findings?
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 17374
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
I will make or purchase LISN. Is there any reason why a LISN designed for 120Vac would be unsuitable to also test battery or wall-wort powered devices?

Standards.  Different ones have different methods, oriented towards evaluating the equipment's output over suitable frequency ranges, or emulating an average environment.  Ground plane or insulating table, high or low impedance, high or low cutoff frequency...

For example, automotive stuff (as mentioned earlier) tends to use smaller inductors with higher cutoff frequencies, because it's more representative of automotive wiring (there's only so many meters of cable in there), and the relatively small overall length of a vehicle means it's unlikely to effectively radiate at low frequencies (< 1MHz?) anyway.

Battery powered stuff of course doesn't need a LISN... conducted emissions are easy to pass. :D

Then again, if it connects to something else via cable -- say, an MP3 player with USB and headphones? -- you need to test it with those in play, using suitable ISNs to verify nothing nasty is coming down the cables.  (USB is kind of a weird, almost silly example; but Ethernet is a fine one.  ISNs are standard for telecom cables, in the range of twisted pair, STP, coax and so on, including POTS, Ethernet and so on.  Obviously(?), one of these ISNs needs to be constructed carefully, to preserve the intended (differential) signal, while tapping off the common mode to the RF system.)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline bson

  • Supporter
  • ****
  • Posts: 1930
  • Country: us
I racked my brains over this and in the end decided on the plain SignalHound SA44B with a set of probes with a +40dB external preamp.  (The Tekbox kit)  The SH Spike software won't do pre-compliance testing since the SA44 lacks quasi-peak detection, and maybe other limitations as well.  It only does EMC PC with the much more capable BB60 real-time analyzer (which is a beauty and triggers my inner geek's cravings even though I have no practical use for it).  I decided on this for a number of reasons:
* EMC PC isn't the real thing, and can't be used to barely inch in under line reliably.  You need to give yourself plenty of margin.
* It's not needed if what you're doing is designing something to be low-emissions and low-interference, low-noise.  Plain old peak/avg modes work just fine for this - you'll still see emissions and be able to track them down easily.  If your product looks good on avg it's still going to look good in a formal test.  (Unless you missed something, but the EMC PC personalities aren't going to help with that anyway.)
* Between the two is sort of a ho hum area.  An EMC PC personality becomes pointless when you drop well below the required standards.
* My hobby projects tend to be more along the lines of "hmm, throwing everything at this how low emissions can I actually make it?" more than "hmm, can I save 12 cents and squeeze in with 0.5% margin by using a cheaper cable with less fancy shielding here?"

For me, for home and hobby use, the answer was meh, not going to spend $500 extra on an EMC kit.  Not worth it.

For commercial use I'd go pick up a Keysight kit with the PC software so get close(r) to the stuff used in the test lab, to shift the odds slightly more in my favor.  Actually, I'd ask what they use then go rent that.
« Last Edit: July 29, 2016, 03:28:42 pm by bson »
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 21236
  • Country: nl
    • NCT Developments
For commercial use I'd go pick up a Keysight kit with the PC software so get close(r) to the stuff used in the test lab, to shift the odds slightly more in my favor.  Actually, I'd ask what they use then go rent that.
That won't be of much help because you'd need the chamber and antennas as well costing major $$$. The goal of doing your own measurements is to verify whether you have improved a circuit enough to pass the test. Passing EMC compliance usually happens by going doing a pre-compliance quick scan first. This results in a graph with problem frequencies which have to be improved. Once the problems are fixed you can go for the big test.

Since measurements can be made relative you don't need calibrated or really expensive equipment at all. Measure the circuit as it is at the problem frequencies using simple magnetic field probes (do note the position and orientation), note the levels from your own analyser and then improve the circuit until the levels have dropped enough dB for the test to pass. I never used the EMC mode on my spectrum analyser for this kind of work.
« Last Edit: July 29, 2016, 05:43:42 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline bson

  • Supporter
  • ****
  • Posts: 1930
  • Country: us
For commercial use I'd go pick up a Keysight kit with the PC software so get close(r) to the stuff used in the test lab, to shift the odds slightly more in my favor.  Actually, I'd ask what they use then go rent that.
That won't be of much help because you'd need the chamber and antennas as well costing major $$$.
Yes Captain Obvious, I know that.  You can still align yourself in setup as far as is viable.  Then when you see their printouts, reports, data, at least you've potentially simplified your life somewhat by using the same equipment.  Or not, but you've definitely not made it any worse and can only benefit from it.  When you go rent a chamber it will have the same setup.  Either way, reducing the number of variables and dependencies on a project is never a bad thing and any engineer should be in the habit of doing so.
« Last Edit: July 30, 2016, 01:10:38 pm by bson »
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 21236
  • Country: nl
    • NCT Developments
I have been through several EMC tests and each time the outcome is (slightly) different for the same product. A lot depends on the technician, the actual setup, cables used, etc. Having variables in the test equipment I use for tests in my own lab is the least of my concern.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline joeqsmith

  • Super Contributor
  • ***
  • Posts: 7830
  • Country: us
I racked my brains over this and in the end decided on the plain SignalHound SA44B with a set of probes with a +40dB external preamp.  (The Tekbox kit)  The SH Spike software won't do pre-compliance testing since the SA44 lacks quasi-peak detection, and maybe other limitations as well.  It only does EMC PC with the much more capable BB60 real-time analyzer (which is a beauty and triggers my inner geek's cravings even though I have no practical use for it).  I decided on this for a number of reasons:
* EMC PC isn't the real thing, and can't be used to barely inch in under line reliably.  You need to give yourself plenty of margin.
* It's not needed if what you're doing is designing something to be low-emissions and low-interference, low-noise.  Plain old peak/avg modes work just fine for this - you'll still see emissions and be able to track them down easily.  If your product looks good on avg it's still going to look good in a formal test.  (Unless you missed something, but the EMC PC personalities aren't going to help with that anyway.)
* Between the two is sort of a ho hum area.  An EMC PC personality becomes pointless when you drop well below the required standards.
* My hobby projects tend to be more along the lines of "hmm, throwing everything at this how low emissions can I actually make it?" more than "hmm, can I save 12 cents and squeeze in with 0.5% margin by using a cheaper cable with less fancy shielding here?"

For me, for home and hobby use, the answer was meh, not going to spend $500 extra on an EMC kit.  Not worth it.

For commercial use I'd go pick up a Keysight kit with the PC software so get close(r) to the stuff used in the test lab, to shift the odds slightly more in my favor.  Actually, I'd ask what they use then go rent that.

How do you like it?   
How electrically robust is your meter?? https://www.youtube.com/channel/UCsK99WXk9VhcghnAauTBsbg
Software, documentation and test reports for the low cost NanoVNA & V2 Plus 4 may be found here:
https://github.com/joeqsmith
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 17374
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
I have been through several EMC tests and each time the outcome is (slightly) different for the same product. A lot depends on the technician, the actual setup, cables used, etc. Having variables in the test equipment I use for tests in my own lab is the least of my concern.

I've heard of (but not yet seen in practice) using a standard test panel for the product: it's strapped to a foam panel, and that arrangement is used for all EMC tests, period.  At least, of the ones where such arrangement is acceptable.  I suppose several emissions standards require different setups, and susceptibility usually requires still others.  But you'd be able to make finitely many panels, for all the standards needed.

Should definitely help with consistency, though, and answer any questions about "well, what should we do with /this/ cable?"

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline Jester

  • Frequent Contributor
  • **
  • Posts: 459
  • Country: ca
In the end I ordered the Siglent SSA3021.

I will provide an update when it arrives. I have the parts for a LISN thanks to Jay_Diddy.

I plan to order a DC blocker, a standalone 10dB attenuator and a few cables and adapters.

Are inexpensive adapters like this acceptable?

 http://www.ebay.com/itm/N-Type-Male-Plug-to-SMA-Female-Jack-RF-Coaxial-Adapter-Converter-Connector-Kits-/152201756624?hash=item236fee7fd0:g:dkUAAOSw0UdXrFus

 http://www.ebay.com/itm/4Pcs-BNC-To-SMA-Type-Male-Female-RF-Connector-Adapter-Test-Converter-Kit-Set-/252477104088?hash=item3ac8cedfd8:g:Hl8AAOSwck5XMVCg

How about this attenuator?
  http://www.ebay.com/itm/Golden-2W-SMA-JK-Male-to-Female-RF-Coaxial-Attenuator-6GHz-50ohm-10dB-Connector-/381607804971?hash=item58d999002b:g:XiQAAOSwXj5XGMZL

Suggestions on cables welcome
 

Offline tautech

  • Super Contributor
  • ***
  • Posts: 21637
  • Country: nz
  • Taupaki Technologies Ltd. NZ Siglent Distributor
    • Taupaki Technologies Ltd.
The SSA3000X series have an inbuilt DC block, I too worried over additional input protection but with feedback from those more experienced with SA's they allayed my concerns.

For additional connector bit and bobs have a look at this well known supplier:
http://www.rfsupplier.com/
They also have an eBay shop.


No affiliation to/with RFSupplier.

Avid Rabid Hobbyist
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf