EMC, how far can you get with DIY? HAM receiver?

[tld]

Hi,

I'm trying to see if I can figure out a bit of a roadmap on how to go about EMC/EMI-compliance, and wondering about how far I can get on a budget.

As I understand it, there's two ways to about assembling supporting documentation behind a declaration of conformity.  One - the preferred - is EMC-testing at a 3rd party lab, which isn't cheap.  The other option is by assembling a technical file by itself, walking through supporting documentation and arguing/proving that EMC/EMI isn't an issue.

First thing I'm not sure about is that I see some resources arguing that this is a choice, while some resources seem to imply that not doing lab-testing is only permissible in some situations, such as when an installation is too large for a test lab, heavily dependent on context (not reproducible in lab), etc.

Does anyone know if you can omit full lab-testing of your supporting documentation is good enough?


I also happened to have a PC-controllable HAM receiver (icom pcr 1000), with AM-sensitivity in the low µV-range (1-2.5 µV), and a fairly accurate dBm-readout.  Given a known impedance (50 Ohm), I could work my way from dBm to dBµV, and given an antenna with a known antenna factor (and cable loss), get to dBµV/m.

Although certainly not something that would pass as a professional setup, I've started to wonder if I could grow on that idea into something passable for pre-compliance testing, or - as part of a larger technical file - good enough to argue CE-compliance?

Uncertainty would certainly be larger than a proper setup, but I have gotten curious if it would be possible to set up an OATS-site (battery-powered for the occasion), run down the tests, and - if results are good enough after taking uncertainty into account - get passable results?

(that is, would it at all be possible to get such a setup calibrated to a point where measurement + uncertainty would still be under limits, thus being able to argue conformity based on those measurements?)


Anything that could be done to improve on the idea to get anywhere?

Note; I'm not saying I think this is an easy solution to a hard problem - I don't - but I'm asking hoping to get an idea of how far from anything usable I might be, and if anyone has any ideas on how to get anywhere.

tld

[w2aew]

I think the best you'll be able to accomplish is to potentially identify gross problems for radiated emissions.  It doesn't address any conducted emissions issues.  The general coverage receiver approach won't give you the selection of filter shapes and bandwidths that are required by the various standards, not to mention the detector types.

Tektronix just hosted a webinar on the Practical Approaches to EMI/EMC Troubleshooting.  While the webinar does focus on some Tektronix equipment and solutions, the general advise given by the presenter for compliance, pre-compliance, and investigatory work is very useful.  Here is a link to the webinar (about an hour long):

[Andreas]

something passable for pre-compliance testing, or - as part of a larger technical file - good enough to argue CE-compliance?

Uncertainty would certainly be larger than a proper setup,

With uncalibrated equipment? forget it.

The problem is that the radiation limits are relatively low.
So any uncertainity will cost you a lot of unnecessary EMC measures like shielding and filtering.

The only thing you can do are comparative measurements. (before and after shielding). To find out wether the next time in lab you will be below the limits.
And even for this the effort for reproducable measurement values is not low. You will need a ground plane in your garden and your next neighbours and radio stations a few miles away.

With best regards

Andreas

[bookaboo]

As with anything "it depends" on the end product.

If your making a mass scale product then there's no way to avoid that test lab.
However your specific question was asked at a CE seminar (more specifically Machinery Directive) that I attended a couple of years back. Strictly legally speaking you are supposed to compile a full technical file for everything you produce, however in practice we were told that it's understood that this is unfeasible.

The thing is your Technical file will only ever be challenged by
a) A very attentive customer, in which case you just have to convince him.
b) In a court of law after something bad has happened.

If you ever read some of the directives, they are more legal than technical documents, they cross refer out to other directives and it can be a nightmare to dot every i and cross every t.
What we were told (by a guy selling his compliance services) is that if you are making a few dozen widgets and you have shown that you took active steps towards compliance such as having a decent technical file, using legit components, decent build quality and common sense then even in the worst case scenario (i.e. court) you will get a favourable hearing.

[KJDS]

You can use the technical file and comparative measurements for minor upgrades, otherwise you really need to be in a real test lab.

Emissions and susceptibility , conducted and radiated need to be accounted for.

[qno]

What kind of product are you designing?

Battery operated or with a powercord?

[mikeselectricstuff]

The only difference between a full certification test (which may well fail) and a 'quick look' session (sometimes called pre-compliance testing) where you go and test stuff, experiment with fixes etc. is the lab's level of liability for the results.
The latter method - just booking some time to do whatever you need can be a very productive use of lab time - you generally get an operator/engineer who has a lot of experience and can quickly identify areas that need attention and suggest fixes. They typically have big assortments of filters, shielding products etc. - I imagine EMC component companies supply these to test houses free, as if something from the kit fixes a problem chances are it will get designed in.
You can try fixes and then get test results that are no different to those that appear in a formal full test report from a product test, and can use these as supporting documentation in a technical file. And you can also test as many things as time permits. e.g. for small products you can test multiple items simultaneously for radiated susceptibility.
Many tests can take quite a while to scan through the frequency range. If you have multiple products, or multiple copies, you can be doing things like applying fixes etc. one one while another is being tested.

A good strategy is to design in multiple approaches to filtering, shielding etc. test the belt-and braces version, and if it passes,  then test the effects of removing stuff, swaping to cheaper filters etc. in order of cost to see how necessary and effective each measure is   
oh, and always leave ESD and surges til last, as these might kill your device, so you can't do any more testing & have wasted the rest of the day. Where possible take plenty of spare units - you don't want to be spending valuable test house time fixing up dodgy protptypes.
Similarly make sure you have enough software funtionality built in to make testing efficient - excersising hardware, giving clear indications of issues for immunity, having the ability to quickly and easily change things that may affect performance - clck rates, drive strengths, PLL frequencies etc.

With uncalibrated equipment? forget it.

And for radiated, outside an anechoic chamber, forget it. There's way too much ambient junk to get meaningful results even with proper EMC test gear.
Conducted is probably more doable, with a LISN and spectrum analyser as frequencies are low and shielding less of an issue.

What kind of product are you designing?

determining which standards are applicable can be at least as hard as any technical challenge. test houses can advise but are not exactly  independent as they will tend to err on the side of including standards & tests