EMC pre-compliance testing 5

[MagicSmoker]

I'd like to be able to do some basic "pre-compliance" testing of high power dc motor drives in house. I'm a power electronics kind of guy, not a compliance engineer, but it looks like all I need to do basic far field (10m) testing is an antenna (preferably calibrated) and a spectrum analyzer.

I haven't found an antenna yet, but I did stumble across this seemingly ideal gadget called the Signal Hound (

http://www.signalhound.com/index.htm)

which turns a laptop into a 4GHz spectrum analyzer.

It *seems* to be a much better solution than buying either a 30 year old HP or a cheap Chinese spectrum analyzer off of ebay, but this might be wishful thinking on my part... Anyone here used one of these Signal Hounds? Even if not, any comments as to why it might or might not be a good choice for what I want to do?

(Oh, and any suggestions for an antenna - biconical or bi-log seem to be used the most - would be welcome, too.)

 

[VEBill]

Given your description of the situation, I'm not sure you should bother trying to make any calibrated measurements. Unless you follow the measurement standards and applicable test procedures exactly, then you'll never know if your measurements are 20dB high or 20dB low.

It might be worth probing to find the largest (in relative terms) emissions in the frequency domain, and then see if you can figure out where they're coming from and try to mitigate them. It's most likely that there would be one band of noise that dominates, so you could probe that with a bit of wire, dangling monopole or a loop, as an antenna.

If you do decide to purchase an EMC antenna, then you need to make sure that it covers the range of frequencies you need to test.

E3 testing can get very complicated and very expensive very fast.

 

[MagicSmoker]

Well that was an unexpected response... Yes, EMC testing is expensive. Yes, the EMC antenna needs to cover the frequency range specified in the test schedule.

Anyway, I went ahead and bought the Signal Hound because the owner/designer confirmed that the software can accommodate antenna correction factors. So now I just need to find an appropriate antenna, a task complicated by the fact that most of our emissions are in the 30-100MHz range but the common biconical EMC antenna tends to perform better in the 100-1000MHz range. It looks like I'll be better off going with a set of dipoles at this point.

The testing methodology is currently in hot debate. The dc drives are for electric vehicle conversions. I'm arguing that some sort of industrial standard should be applied; the Europeans are (weakly) arguing that I need to use CISPR 12 (which technically only applies to spark-ignited motors). I cannot exaggerate just how little fun this process is, but at least being able to do a good chunk of it in house before spending the big bucks at an official lab might make it possible, if not palatable.

 

[SparkyLarks]

From a European point of view I would say the Automotive EMC Directive ,2004/104/EC or the EMC directive 2004/108/EC must be complied with.

As far as I can see if the Drive system is considered an Electronic sub assembly, and I would think it is then 104 is the guy. But I'm not familiar with 2004/104/EC at all

http://www.acero.ro/Directiva_2004_104_automotive_eng.pdf

AS for the testing i've never had to do it myself. I am normally able to prove conformity by use of harmonised standards and good installation practice. But unless your electric vehicles can be classed as a fixed installation that route isn't open to you.

The approach you have taken is probably a good one. See what EMC disturbances are then, mitigate them as best you can and then call in an expert to help you with the final certification.

 

[MagicSmoker]

We just sell the drive as a standalone product, and since it can be used for more than just electric vehicles we're going for a CE mark under one of the industrial equipment directives. The vehicle itself is getting tested as a whole for "type approval" so this neatly sidesteps testing the drive itself as an ESA under 2004/104/EC.

That said, I am really most interested in hearing from anyone else who has bought the Signal Hound spectrum analyzer, particularly for ad-hoc EMC testing like I'm planning to do. I'd also appreciate any tips on where I might get antennas for a reasonable price. And any anecdotes about the worth/non-worth of doing pre-compliance testing in-house would also be welcome. Thanks.

 

[zappedagain]

I've done testing with a (roof top) TV antenna before; it is clunky but it works. That doesn't quite cover your frequency range but you get the idea.

When I test I only need relative measurements, such as when my test house (luckily I can pretest fairly inexpensively) finds a frequency spike of concern. I can then go back to my lab and measure to see if I can knock it down enough so that I know I'll pass when I go back for the full test.

The sweep rate on the hound seems a bit slow, but you get what you pay for. You'll catch any CW emissions but modulated emissions may be difficult to detect. You'll have to lock on the carrier frequency and use the demodulator tools. If you have to do this a lot that is where the dedicated spectrum analyzers start to show their worth.

Z

 

[biff44]

I know what you mean. Getting a calibrated antenna will set you back thousands of $. Go with one of these Discone antennas:

http://cgi.ebay.com/Base-VHF-UHF-Di...ectronics_Radio_Equipment&hash=item35afc0d2b3

I use these fairly frequently to scan for unwanted signals. The big ones that pop up above the noise...I then take a tunable monolpole whip over a ground metal plane or a dipole, measure that one spike, and tune the whip for maximum signal, and figure out the approximate signal strength assumeing the whip or dipole is near its theoretical gain value.

http://www.MaguffinMicrowave.com

Maguffin Microwave wireless design consulting

 

[biff44]

BTW, you can probably find that antenna at many ham stores or at plenty of USA places online.

http://www.MaguffinMicrowave.com

Maguffin Microwave wireless design consulting

 

[VEBill]

For approximately zero dollars, one could use 'any old wire' to find the frequencies of interest (stand a bit closer if req'd), then cut coat-hanger(s) into tuned monopole(s) (or dipoles) for each frequency of interest.

Making simple antennas is about 'a million times' less complicated than designing the product-under-test.

 

[biff44]

Yes, but you will never know which ones are big and which ones are small, because the wire will resonate/anti resonate all over the place. For precompliance, you really want to "sniff" with something that at least attempts to be broadband.

http://www.MaguffinMicrowave.com

Maguffin Microwave wireless design consulting

 

[VEBill]

One of my assumptions is, "It's most likely that there would be one band of noise that dominates..."

If that assumption is true, then a simple wire probe will find the peak frequency (or frequency band) very easily.

If that assumption is not true, then the OP can make a couple more dangling bits of wire (of varying lengths) to ensure that the probe doesn't have any sharp peaks or nulls that could fool the unwary. Such sharp peaks and nulls are unlikely for a 'short' probe or a 'small' loop.

Consider also that my suggestion takes less time than filling out the paperwork to order a commercial antenna.

Of course one needs to be wary; but that applies even if you use a commercial antenna.

 

[MagicSmoker]

Thanks for the comments, guys.

However, what I really want to do is make AMPLITUDE measurements with reasonable accuracy over a wide frequency range, not identify what frequencies are the main culprits. I really don't want to cut up pieces of wire and hope the results are meaningful. I'd really like suggestions on where to get calibrated antennas, preferably gently used.

Besides, I already mentioned that most of our emissions are in the 30-100MHz range; now I need to quantify how various tweaks to the hardware and installation affect the amplitude of the emissions. In other words, I've already identified the problem frequencies. Specifically:

Freq measured result limits
(MHz) dB(?V)/m dB(?V)/m
30 51 32
32 53 32
34 57 32
36 57 32
40 41 32
42 36 32
44 32 32
50 40 32
53 58 32
58 39 32
65 38 32
68 34 32
87 40 33
90 45 33
92 47 33
94 51 34
112 26 35

zappedagain - it doesn't look like the Signal Hound's slow scan rate is going to be much of a hindrance here, does it?

biff44 - Thanks for the Discone suggestion! A 10:1 bandwidth sounds excellent here. I'd almost be willing to roll my own, designing for, say, a minimum frequency of 20MHz. However, one potential downside: don't Discones like to be far away from ground planes? Another one is that it might be difficult to correlate the results from the Discone with an "approved" antenna design like a dipole, biconical, bi-log, etc.

VE1BLL - my rationale for getting a proper antenna is that it is the thing mot likely to be non-linear and by the widest margin. What if it says I am persistently 10dB too high at some frequency when, in fact, I am 10dB within the limit? That's not just useless, it is actively bad, since it might result in me spending lots of time trying to fix a problem that doesn't exist.

 

[VEBill]

Sorry, I had read too much into the *pre-* in "pre-compliance testing". I didn't realized that you were that far along, and that you already had (calibrated) measurements of the UUT E3 failures on a frequency by frequency basis.

So hit eBay, or Google your way to any of the many "used test equipment" vendors.

Of course, if you reproduce the test setup, then you could use those lab test results to "calibrate" your measurements.

 

[MagicSmoker]

VE1BLL - Sorry I wasn't clearer about what I needed in the first place. I've bought a few things off of ebay with mixed results. Same with a couple of the used equipment vultures. We bought an ancient "as-is" Quad IIc pick-n-place from one that turned out to be the deal of the century ($2500, and it worked right away) but other stuff has been much more hit-or-miss, emphasis on the miss. Trying to duplicate what the test laboratory uses is not even remotely feasible. Just their spectrum analyzer (a Rhode & Schwartz model) is $38K used.

Fortunately, I think the Signal Hound spectrum analyzer is going to work just fine, but there are a bewildering array of antennas to choose from and none of the manufacturers seem to want to tell you any prices unless you email/call them. Tedious.

BTW - the previous results are with a 240V nominal DC supply feeding the drive while it delivers 200Arms into the motor. Just to put things into perspective.

 

[MagicSmoker]

biff44 - I bought a "25-1300MHz" scanner Discone similar to the one you suggested from an ebay seller here in the US. $57.20 shipped! At that price it's stupid to not buy one and see how it works. It's a "JTD1" which seems to be rather popular. The cone elements aren't really long enough for 25MHz, but at this price who cares?

I also found an Emco 3121C calbrated dipole set for $999. Sounds like a pretty good deal. I can use that to check the discone then maybe sell it to the next poor schmuck in the same predicament as me.

Every used-equipment dealer had "request quote" or "call for price" on their antenna listings. I don't waste my time on those shenanigans.

 

[MagicSmoker]

I've been giving this calibrated antenna problem some thought this weekend and it occurred to me that it should be possible to null out the effects of antenna (etc.) non-linearity with a tracking generator feeding an identical antenna as used on the spectrum analyzer's input.

Indeed, I plugged this idea into my google machine and it does not appear to be a novel one (which is always a relief to me - I hate it when I come up with a novel idea... it usually means I'm wrong). I'm not sure, but I think that this is not acceptable for actually "certifying" an antenna as calibrated, but it should be close enough for pre-compliance work.

So, my nominal plan as of this moment (subject to change, of course) is to get a second Discone antenna and the tracking generator option for the signal hound then have at it. Sound like a reasonable plan?

 

[VEBill]

Have you read about the Three Antenna Method?


Also, as I hinted before, if you can keep the unmodified unit on hand (the prototype unit that gave you all those readings), then you should be able to make relative comparisons (at each frequency) between your first prototype and the second (with your E3 mitigations incorporated). Then the antenna calibration wouldn't matter, so long as you ensure it's stable while you switch UUTs. If you have only one prototype, then disregard.

Good luck.

 

[biff44]

Yes, I have a tuned dipole like that too. It is a valid way to do things. If I had perhaps $2k+ I would instead go for a calibrated broadband antenna that did not need to be physically tuned, such as this style:

http://www.ahsystems.com/catalog/bilogical.php

That way you can look at the full frequency sweep at a glance and see if you are in/out of spec.

You will need something to hold these antennas (preferrably a non condutive tripod) and a big low reflection area to make the tests (like a back yard, or a big warehouse without any metal nearby.

One good thing about the dipole type antennas is that they are physically small, so you can use them in a smaller lab. Also, easier to hold them up, and rotate for dual polarization tests. Just a pain to tune them for each frequency!

http://www.MaguffinMicrowave.com

Maguffin Microwave wireless design consulting

 

[VEBill]

"...a back yard..."

Usually such testing in performed in a shielded room. In this case, there are two reasons to be very careful about extensive outdoor testing.

1) The UUT is emitting 'illegal' amounts of EMI.

2) The frequencies listed (30 to 112 MHz) include public service, pagers, ham radio, VHF TV, and FM broadcasting. Some of these would be random enough (on/off) that you could not subtract a stable baseline.

I'm not claiming that one cannot perform some minimal testing outdoors, just take care to avoid the possible problems.

 

[MagicSmoker]

Thanks for the reply, biff44. I actually gave that specific AH Systems antenna a good long look but didn't pursue it any further because of the "request price quote" nonsense. If it is around $2k, though, then it's certainly within what we would consider reasonable for our purposes.

I just got a response from Signal Hound and they say that it is possible to use their companion tracking generator to null out the response of the antenna by feeding a second "identical" antenna with it. So, unless anyone has any good reasons why this is a bad idea I'm going to pursue it. If nothing else I can use it as a network analyzer...

Yes, VE1BLL, it does appear that the 3 Antenna Method would give better results, but it's not clear to me that it would be worth the hassle. I'm interested in setting up an ad hoc/impromptu EMC pre-compliance testing scheme, not a metrology lab.