I'm building a chassis out of anodized aluminum panels. It occurs to me that, if these panels are nicely insulated from each other, they will do a rather bad job of blocking EMI.
Now, this wouldn't ordinarily concern me much, but I happen to have a project which doesn't simply produce an annoying amount of EMI, it produces enough that, if the chassis were steel instead of aluminum, it would get burning hot to the touch in spots. That's not good.
So the question is, does anyone know if this will be a problem? Does anyone have any anecdotes concerning similar construction?
If it's a problem, I'll just grind down the anodize around the flanges and hope that helps.
Tim
--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms
It's a problem, all right. One time when I had a persistent 80 MHz spur in a heterodyne interferometer, I eventually traced it to the anodized bottom cover of a Mini Circuits power amp not making good contact to the rest of the case--even though the anodizing was removed around the four countersunk screws holding it on.
Thirty seconds' work with a Dremel got me a 50 dB spur reduction.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net
It can and does help. Grind down the anodization. If you're having the panels anodized have the anodizing house mask off the mating flanges next time -- it's routinely done by at least some of them, and AFAIK it's not prohibitively difficult to do.
Yes, they do a very poor job as the wall elements (as well as the floor and roof panels) of the chassis are not all electrically integrated together, as would be required to establish a Faraday cage.
. You need to get down to bare Al on the mating faces, and use a good silver filled epoxy to get a reliable RFI/EMI seal.
Alternatively, if you're starting with metal and having a shop finish it, alodining will preserve the conductivity and you'll only take a small hit in shielding effectiveness. (In the ballpark of 5dB, whereas anodizing will often lose 40dB+.)
Hot to the touch? Are you using open core inductors? I think this will cause some EMI issues. With lots of Conducted noise. Closed core inductors (gapped) and/or toroids would be best. Will it be used next to other sensitive equipment?
One would want a gasket in a "lid" location or other service access. The main body of the chassis, however, will want to be as tightly physically bound as possible, so electrical integration of all the non-service oriented panels should come from a more direct method. A gasket would diminish, generally, the physical rigidity of the panels that make up the chassis box, or require an size increase to maintain it. A direct bound method would yield both rigidity and Faraday cage level EMI "sealing".
Maybe grinding is not the answer. Perhaps an "L" shaped gasket on the inside corners will make a waveguide operating above cutoff for rather high attenuation?
Well, not in theory. But I think the secondary is throwing off a lot of stray field anyway, plus the leads to it.
formatting link
Basic layout: barely visible heatsink is FWB running from AC line (white zip cord, coming from an isolation transformer ATM). Runs electrolytics (2000uF, hardly visible), yielding 160VDC. A load of random film caps create a split supply for the output. Inverter chops +/-80V into the toroidial output transformer (to the tune of 20-50kHz, 160Vp-p, ~200ns edges), which has 1/4" copper tubing for the secondary. Secondary is in series with total 10uF caps and the work coil (3uH coil shown).
During one test, I had the output network stuff sitting on a steel plate, which got burning hot in places, particularly under the capacitors and transformer. Hence, "not just annoying EMI, burning hot EMI". Aluminum will of course do better than steel in that regard.
Nothing special, but I have two immediate concerns: when the inverter is running near the DC-DC converter's frequency, they interact, producing quirky behavior (e.g. whistling). The DC-DC converter is the UC3842-based circuit along the front side of the big PCB, it supplies the gate drives. I'm not sure which one is pulling; if it's the DC-DC, oscillating gate drive supply voltage would be very bad. If it's the inverter oscillator, that's not as bad, but still undesirable. I have filtering between them, so I'm not entirely sure how it's pulling; maybe I missed a ground loop.
Also, I can't see much on the 'scope at less than 100mV/div, because the inverter is throwing spikes everywhere. That's an induced noise issue: I can short the probe to its ground clip and the noise remains, but amplitude drops when I hold the ground lead tight to the probe body. I've considered snubbing the inverter, but it has to accommodate possible hard switching as well, which makes the snubbers awfully complicated (dV/dt and dI/dt, bleh). So I'm hoping I can get away with it the way it is.
Tim
--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms
Spot heating can be a problem. This is field-induced, so identifying the 'spots' is important in simple avoidance through layout and location of fields with respect to affected hardware. A steel external case will work best, but local hardware for mounting, field shaping and flux-shorting can be of diamagnetic or paramagnetic materials.
Induction heaters are affected by EMC standards, though these may not be the ones you're used to working with. Conducted standards will apply.
Induction heaters are generally not expected to be squeezed into small cases. Don't be unnecessarily or arbitrarily silly in your package design.
Say, how can I measure those at home, and what am I looking for? Is less than so-and-so milivolts okay? It varies with frequency, of course; I recall a plot of dB vs. frequency. But that's dB relative to what, line voltage? Line power?
Yep, I have a good 6 x 11 x 13" box for this 1-2kVA project. The output network is at least 2" from any nearby wall (except where it has to go through one side). It should be okay, as long as the shielding works.
Tim
--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms
Measurement methods aim at standardization and repeatability, where ever they are carried out. For conducted measurements, you'd need an LISN to start. Measurements at spot frequencies usually require a frequency-selective voltmeter or some form of spectrum analysis, but the fundamentals and lower harmonics are often large enough to see on a scope - especially if they approach the limits of the induction heater you're dealing with.
The LISN is intended to provide a definable source impedance, while assisting in attenuating the low frequency power signal. Conducted emission limits in the conventional 150KHz-30MHz range can be in the tens of microvolts range - rather difficult to pick off residual fundamentals that are still in the volt range.
~From
formatting link
(annexJ5)
For an industrial environment with dedicated independant supply distribution typical of a heavy engineering EN50081-2 (generic) or EN55011 ("ISM" equipment) are likely to be the most relevant emissions standards.
Light industrial environments that share distribution with office or consumer facilities EN50081-1 (generic) or EN55014 (electrical appliances) are likely to be the most relevant emissions standards.
Line harmonic emmision limits for both environments are outlined in EN60555-2 (EN 61000-3-2).
An emissions limit chart, courtesy of Schaffner, is temporarily located at:
formatting link
Note that 55011 tracks the field strength as well as conducted emissions down to 9KHz.
Conducted dBuV in a 50ohm sensor can be converted into conducted uV using the 20xlogE relationship and into conducted uA by means of Ohm's law. The field strength axis refers to a 2 meter diameter loop antenna as a sensor.
ISM apparatus which comply with EN55011, such as plastic welders or sealers, induction heaters, microwave cookers or dryers, can still have extremely intense emissions in certain radio frequency bands, even so far as to create personnel safety hazards. This issue is covered by the relevent safety standards: these normally contend with field strength and human body absorption issues above 100khZ. (oet56 and oet65 in the US).
"The use of the standard IEC 61508 is strongly recommended when dealing with the functional safety of any electrical/electronic/programmable electronic system."
Silver filled epoxy does a "broadbanded" job of sealing it all up.
EPOTEK "H20E" IIRC.
formatting link
ALL the chip makers use it to bond chip dies to their heatsinks.
I have seen good results with both a small bead applied internally (inside the case) at the seam in question, to actually being integrated into the seam while it is being assembled (between the mating faces). It's like grouting! :-) (not in texture though, that is creamy).
The bead works fine. We dremel tooled the corner to make two conductive, fresh metal faces, and put a bead of the epoxy on that. 4 Hrs at 80C, reassemble, and no more spurs!
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.