DIY EMI box

[Bigman]

Hi,
I am teaching myself in electronics and recently I have build a basic transimpedance amplifier. But when I am measuring ths output-signal with my oscilloscope, I also catch a lot of ambient noise/signals. So, I have build a quick-and-dirty box, which should protect my circuit under test from such external influences. As materials I have chosen tinplate (against E-fields) and layed it out with copper (against H-fields). Also it's working you can see from the picture the whole assembly is very cheesy, which is the reason why I want to build a better version of it:

The main design-improvements I have in mind already: to avoid any gaps between the lid and the bottom, either I will use finger stock or I will design a kind of flange so I can screw both parts together;

Where I need your advice:
- when the box is closed, I cannot use my normal probes from the oscilloscope anymore. So, I placed BNC-ports on the box-wall, and soldered a resistor and adjustable capacitor in parallel on it. From there, I am then running normal wires to the circuit under test (well, in the picture it's just the simple battery).

I am pretty sure, many people already have designed their own EMI-box and perhaps they could share some (low-cost) hints.

Thanks!

[SeanB]

Biscuit box, and as a side effect you get to eat the contents.  Metal Box makes them locally, I know a retired sales manager from there, and they do do some pretty big ones.

[Bigman]

Biscuit box, and as a side effect you get to eat the contents.  Metal Box makes them locally, I know a retired sales manager from there, and they do do some pretty big ones.

Indeed the biscuit box also crossed my mind several times. Most likely, when I find one which fits my dimensional needs I will go with one. If not, or the results are still not satisfying, I will go with the the finger stock/flanges.

 However, still the problem with the probes remain. Let me run a test-example with my current setting to demonstrate one of the major external noise/signal effects I would like to overcome (I will post pictures in a few minutes).

[Isad]

Nivea cream box or any round metal box would do
and the nivea one would even be a bit tiny wich idk if that
fits the battery.

[Bigman]

Okay, as a circuit under test, I have just placed a virtual-ground driver into my EMI-box. I use this driver to split a 3.8 battery voltage into 70mV and 3.63mV (once I needed this for a dual-supply OpAmp). The whole set-up and how the wires are running can be seen from the picture full-setup.png.
When the box is open, as expected, I am getting a lot of noise (see pic. EMI-box_open.png).
When the box is closed, it already looks much better. But STILL, there is an interfereing 50Hz signal (see pic. EMI-box-closed.png). BTM: I am living in Germany, and our power-supply runs on 50Hz.
I assume, I am "catching" that 50Hz external signal by the coaxial-cable running from my EMI-box to my oscilloscope. Indeed, I soldered the cable and the BNC-ports together by myself.

So, in general I am wondering: "since you cannot use your normal probes with such a box, how should the probes be substituted?

Thanks again!

[SeanB]

Most pick up is from the cable and the scope front end. Spikes are conducted noise from the mains wiring, either arriving as common mode voltage on the mains cable or radiated from the wiring in the room. You would need an isolating transformer for the scope ( one that allows the mains earth to be connected to a separate low resistance earth mat via a thick wire or strap) and some dual screened coax as well, so as to reduce this further. Metal box is probably giving over 50dB of attenuation already.

[Bigman]

You would need an isolating transformer for the scope ( one that allows the mains earth to be connected to a separate low resistance earth mat via a thick wire or strap) and some dual screened coax as well, so as to reduce this further.

Those are very good hints, I think I can work with.
First - as you recommended - I will check where I can get dual screend coax (and what else my shop has available regarding better shielded cables).
Second: if the new coax cable still don't show significant improvements, then I will look for the isolating transformer (I assume such a device will be more expensive than the coax cables). Indeed, I have never worked with such a transformer. I believe you mean something like this http://en.wikipedia.org/wiki/Isolation_transformer

[SeanB]

Cheaper initially would be a mains filter that also has an earth lead choke, or use a 3 phase filter unit, and feed the incoming mains wires ( As it is Germany and you use a socket that does not prevent neutral from being a phase voltage but considers it as an active power line while most appliances are only going to switch one side of the mains) through 2 of the sections, with the PE of the scope connected to the output of the third, input of the third being connected to the case and incoming mains earth.

That will reduce the mains noise on the incoming side, and is almost as effective as an isolating transformer in this limited case. Cheap way to see if the noise is conducted noise or induced into the coax cabling.

[Bigman]

Cheaper initially would be a mains filter that also has an earth lead choke, or use a 3 phase filter unit, and feed the incoming mains wires ( As it is Germany and you use a socket that does not prevent neutral from being a phase voltage but considers it as an active power line while most appliances are only going to switch one side of the mains) through 2 of the sections, with the PE of the scope connected to the output of the third, input of the third being connected to the case and incoming mains earth.

That will reduce the mains noise on the incoming side, and is almost as effective as an isolating transformer in this limited case. Cheap way to see if the noise is conducted noise or induced into the coax cabling.

wow ... I highly appreciate your kind answer! I quickly checked what would be available from my local shop. Indeed, they have filters starting from 2.- Euros up to >1000.- Euros (!). Pls. allow me to counter-check with you: are you speaking about filters as described in the links below (they cost around 5.- to 12.- Euros):

Those are called "main filter":
http://www.produktinfo.conrad.com/datenblaetter/500000-524999/521315-da-01-en-NETZFILTER_1A_250V_ANSCHLUSS_OBEN.pdf
http://www.produktinfo.conrad.com/datenblaetter/525000-549999/530184-da-01-ml-NETZEINGANGSFILTER_3EEJ1_de_en.pdf

This is called "noise filter" (but 1phase ... I still have to read your thread a few more times to understand your 3phase explanation ... keep in mind: I am a newbie :-) ... but what I have already seen is that 3phase filter costs over 200,- Euro ):
http://www.produktinfo.conrad.com/datenblaetter/375000-399999/398717-da-01-en-1_PHASE_FILTER_40_C_20A_FN_2020_20_06.pdf

But when I look on the spec of all those filters, one point is puzzling me. They show charts regarding the attenuation. This attentuation (which I assume is the "damping" of my unwanted noise-signal), is getting very close to 0dB at frequencies below 100k ... well, my unwanted noise-signal frequency is just only 50Hz. Will they really show than any effect?

Thanks again a million time!!!

[SeanB]

Yes, though they seemingly do not supply filters with an earth choke, so while it will clean up the mains you still might get some noise coming up the grounding wire which is connected to the front panel input BNC on the scope.

[Bigman]

For sure, I will buy and test the idea of using a filter for the power supply (even I still haven't found one with an earth choke. Indeed, there is e.g. the Schaffner 9244e, but no dealer has it on stock or a minimum purchase quantity is requested).

While seaching for a filter, I was wondering: shouldn't such a device not already be equipped with appropriate power-supply filter(s). So I opened my oscilloscope (and said bye-bye to the warranty :-) ).
The black block on the right is the connector to the main power-supply. On the PCB where it's mounted on, there are quite some coils which may be filters, already 

[SeanB]

Looks like you will not find one, they seem to be rare these days. Yes there is a filter in the power supply, to keep the stuff inside out, but adding extra one outside to keep the outside in does not hurt at all.

[cellularmitosis]

Looks like you will not find one, they seem to be rare these days. Yes there is a filter in the power supply, to keep the stuff inside out, but adding extra one outside to keep the outside in does not hurt at all.

On that topic, I have a handful of these mains filter boards left over after dissecting a bunch of microwave ovens a while back.

http://www.appliancespares.co.za/6355/kelvinator%20microwave%20oven%20noise%20filter.aspx

I was thinking about putting a few of them in a mains junction box to add additional external filtering to my scope.  Is there any harm in ganging multiple of these filters in series?  (e.g., would that start to cause mains phase lag and somehow cause problems or strange power factor issues?)

[T3sl4co1l]

Harm?  Doubtful.

Benefit?  Also doubtful.

Depending on which side of that network goes to ground, it'll either present a high or low impedance to conducted RF.  If a high impedance, some radiation will still be emitted from the scope's line cord (effectively a 1/4 wave whip antenna).  If a low impedance, the same is still true, but the scope end may be the "free" end of that antenna (which probably means more conducted noise up over the probes, which is bad).

Have you determined that the scope is, in fact, noisy enough to warrant action?  Are you performing sensitive measurements that the scope's own noise is interfering with?  Have you been able to confirm if the noise can even be filtered (is it getting into the scope's front end, internally, or is it happening at the connector or probe level)?  Can you perform those measurements with shielded BNC cables instead?

On the occasions I need better filtering, I just wrap the probe cable around a humongous ferrite bead.  This doesn't usually eliminate anything, anyway, but it does shift resonances to a lower frequency where it's painfully obvious that it's something about my circuit.

Tim

[Bigman]

The double shielded coax-cable together with the BNC-connectors (which are especially for this cable) just arrived. Compared to the ordinary coax-cable, these double-shielded once are really, really thick, which also makes them very stubborn to handel. But they do a great job. My initial problem with the EMI is gone!
Conclusion: those double shielded coax-cables and especially their BNC-connectors are relative expensive - but worth the money.   

Thank for all your contributions!   

Supplement:
I just came across those two videos from Dave (which are pretty good explanations regarding the noise I had, incl. some counter-measures):