EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: rhb on April 09, 2018, 01:16:59 pm
-
I've been having problems with EMI trying to make some low level measurements. Using a small loop of copper wire on the end of a scope probe I tracked down and eliminated the noise coming from my Z400 PSUs by placing a 4" square of 1/8" galvanized steel screen over the PS fan opening.
As the general noise level goes down I can see individual sources more clearly. I was quite stunned when I probed around my Instek MSO-2204EA. The front of the DSO is clean, but around the back and underneath there is a lot of EMI radiated.
The photos show the loop probe I made and the noise pickup at 1x and 1 mV/Div. Good Will says there is nothing to be done about it. Of course, that really means they won't do anything about it. But I shall.
My RSP2 from SDRplay has a sprayed shielding applied to the plastic. Spraying the inside of the back cover is easy. It's the labor of getting the front panel off that I don't relish doing. Does anyone have any experience resolving a similar problem? As painful as a complete disassembly would be, that seems the best option for reducing the noise from the scope.
It would be really nice if just spraying the back cover solved the problem.
I've got ferrite clamp on chokes on order. I still haven't sorted out how the noise is getting into my measurement setup, but pick up and reradiation by power cables seems a possibility, hence the ferrite choke order.
The loop of copper wire is not a proper EMI probe, but I am quite amazed how well it works. Especially as it took only a couple of minutes to make. It's not even noticeable on the time domain trace, but on the FFT I can see a small peak appear and disappear if I rotate the loop near the MSO LA port. On the Z400s it had to be within an inch of the fan opening to show anything on the DSO, so it does a good job of localizing the source. After I installed the screens I went all around the Z400s and could find no additional emissions.
-
My oscilloscope's own PSU causes a nice big mess on the power cord, big enough that the first time I noticed it (then not yet knowing where it was coming from) I was probing well on the front side of scope. I ended up simply laying a steel |_| beam on top of the cord (i.e. cord running under that "house"), not even grounding it. Even if it didn't cover all of it, that was enough to drop the noise below floor (as observed on the front side working area). I tried to look for power cords with braid/shield inside, but couldn't find any for sale. (Afaik, the scope's innards are inside steel box structures, so in my case there would be no benefit from more shielding on the case. Would have to actually have proper EMI filter on the IEC connector or such.)
So, if you suspect radiation from the power cord (EDIT: or the lead between PSU and device), perhaps simply try wrapping some of it in foil. That might not block magnetic field like my 1mm+ steel thing, but who knows, maybe it will help some by blocking the electric field.
-
The most effective is to kill it at the source if you can find it?
You'll have to sniff around with your small inductive probe to locate it.
Sorry I can't be more specific. :-//
-
I'm pretty sure the dominant source is an unshielded transformer based on where the noise is strongest. Time to open it up and study the problem.
I've got steel tubing large enough to drop the power cord through. It's got a 3/'8" wall thickness, so it is rather heavy, but certainly a good test. However my ferrite choke cores came and putting one of those on the power cord had no discernible effect on the noise. Nor did one on the power lead from the linear bench PS. I do need to clean up the bench PS output though. It's OK for most things, but not when you're trying to measure below 1 mV.
Depending on how things go, I may make a filtered and shielded power distribution system for the bench with minimal length power cords. But I'll need to be certain that I have things where I want them.
-
If your SMPS has perforated shields, it's possible that there are emitters which are close to the surface. The shielding effectiveness is reduced if this happens.
Can you see the frequency of the burst content? And what's the frequency (or period) of the bursts themselves.
If you have any space between the SMPS and the plastic case, try positioning a parallel piece of screen (copper or brass, aluminum can have oxide problems) parallel to and short distance away from the SMPS shield. Bond it well to the SMPS shield at multiple places.
As a simpler test, you could lay a sheet of aluminum foil under the scope bottom and intially, just ground it to your BNC jacks. For low inductance, wrap the foil to reach the jacks as a wire of any length will reduce any improvement. If the noise get worse from this "image plane", it means that you just pumped some return currents into the analog front end ground via the jacks. Then, you should try bonding to your chassis / SMPS shield instead.
Whatever you do, respect your SMPS and line voltage (or greater).
-
I have a bit more data.
I've got strong conducted EMI from the 4 ft LED lamps. I have filters on order.
The DSO SMPS problem continues to be rather mysterious. With the supply disconnected from the shielded, twisted pair power cable to the DUT, no noise. Connect the supply to the cable, lots of SMPS noise even when the bench PS is not plugged in.
My linear bench supply appears to be an extremely effective antenna. Based on the preceding, I connected the PS to the DSO via a banana to BNC connector and a 50 ohm cable. No matter where I move the PS, the signal on the DSO does not change. I will repeat tomorrow with a longer BNC cable.
There also appears to be crosstalk between DSO channels, but I need to come up with a more refined test. But noise on channel 1 appears on channel 2 in some circumstances.