H field and E field probe
I am a power electronics hardware design engineer presently working on EMI compliance of my product. I have been using a spectrum analyzer from GW-instek (GSP-9300). I have a set of near field probes that came with the product . Following is the link to their solution.
http://www.gwinstek.com/upload/media...t_solution.pdf
The loop antenna being the H field probe ,my guess was in the near field it would only be picking up di/dt , current changes. The E field probe with the sharp tip should pick up Voltage changes.
Surprisingly,after making sure current flow in my system is minimum by reducing duty cycle I did not observe any appreciable change in the emission levels that I was capturing. I compared it to the voltage probe the levels were similar. I do not want to bias your judgement but I have a feeling the H field probe is not shielded against E field and hence the observation.
My idea is that during near field probing I should be able to differentiate between an E-field and H-field source given proper probes. Please share your views.
Without a schematic and information about probe position, it's hard to determine if your expectation of low H-field is substantiated. You also didn't mention the observed frequency range. At higher frequencies, there will be a coupling of E- and H-field by nature.
I however agree with your consideration that the H-probe is most likely sensitive to E-fields to some degree.
Both small loop and stub antenna will give similar near field signals from harmonics of SMPS pulse currents.
The small loop may be easier to locate the source of emissions, but often the conducted emissions become radiated via the interface cables, so these should be well shielded and use large ferrite donuts or clamped C shells around interface cables to reduce those effects. Most like sources are the transformer and traces to switching devices.
Far field measurements use biconical wideband antenna for E field and large loop for H fields at LF only.
If the duty cycle of the full-bridge system switching at 100 Khz be made Zero (overlap of non-complementary gates). Then there should not be any H-field produced. The small amount of H-field generated due to charging of parasitics, should at least reflect in the spectrum analyzer as decreased H-field as I move towards no-load from full-load.
Restating my question. If I am within the bandwidth of my E-field and H-field probe, create a situation where we have say we have high dv/dt ( an oscillator ) should the near field E probe pick up more amplitude than the H field probe?
If the H field probe is picking up change of flux from my system, shouldn't it be directional ( affected by change in angle of probe at the same point current flow being on a 2 dimensional plane)
My system however is inside a 3 mm thick continuous aluminium enclosure. The readings are being taken above the gate-driver.The frequencies of interest are 7-205 Mhz.
What do you exactly mean with "full bridge duty cycle of zero"? I would expect one high side and one low side switch full on, but apparently you mean something different, e.g. unipolar pwm with both half bridge switching in phase?
Gate peak currents will be high as long as the bridge keeps switching, currents generated by charging Cds of output transistor are rarely small.
Yes, it should be directional. But instead of the assumed "point current flow" you have most likely multiple extended conductors contributing to the overall H-field. It's often impossible to identify a single source.
I personally prefer very small loops (e.g. 3 to 5 mm diameter) to soldered to a coaxial cable end to locate regions of highest dI/dt.
H field loop gap will be most sensitive when gap is smaller than diameter. WHen I did not have loop sensors I made my own using scope probe shorted to ground clip in a loop.
Both stub antenna and loops are directional with notches in response along axis of antenna. Since then loop antenna is flat , its minimum response is in the same plane in all directions as the loop and better for current pulse induction than stray capacitive voltage coupling into a stub antenna.
When looking for emission problems, view all mechanical slots as radiating antenna and conductors as well. Using good ground plane connections reduces voltage emissions but if not careful in ground spike current paths can raise conducted emissions.
Without full pictures and details, it's hard to guess what you are trying to achieve,
I recall in the early80's measuring ingress into cable TV coax in a back lane beside the tracks. whenever a locomotive went by with high current traction motors , there was considerable noise in the TV spectrum similiar to your region of interest. Coax quality and poor distributed ground resistance losses were the reasons. Check your grounds and cable shield quality.
FvM and SunnySkyguy,deeply gratified,your posts were insightful. Its a pity I could not share any pictures or schematics with you for I.P reasons. However I would come up with some way of sharing an equivalent situation to give you a clearer picture.
To give you a scheme of things I am doing pre-compliance testing for a 1-kw SMPS encased in a 3 mm aluminium casing.The near-field H probe I am using, is used to take readings at various points outside this Faraday cage. Incidentally I ran some calculations and found that 1 mm mild steel has far better near field shielding capabilities. Do you have any practical cases which you could testify to?
I won't expect significant H-field outside the enclosure in the stated 7 - 200 MHz frequency range. It's true that steel is more effective in shielding residual low frequency (e.g. up to some 10 or may be hundred kHz) magnetic fields. MHz magnetic fields will be quantitatively absorbed by eddy currents in an aluminium case.
MHz interferences will be mainly conducted by external cables (power, load, control) and radiated through enclosure voids.
I expect your radiated signals are due to conducted input pulse currents between box and "ground plane" , or it is leakage in the case lid seal in the box itself. Perhaps the Alum is anodized with poor RF contact.
RF Foil tape with conductive glue is often used to confirm lid leakage. BRaid shield gauze is often used to wrap around input cable to case and ground and plane to determine if this is the cause. Then investigate line filter pulse current suppression design.
Is this for compliance to Class A, B or lower (Tempest) emissions?
Pls confirm all the above.