I'd like to check the area near a sensitive circuit for airborne RF. I was thinking of attaching an antenna to a bnc cable and running it to a spectrum analyzer. But what sort of probe/antenna should I use? I would think I would need one that had a flat response through the range of frequencies I'm interested in. True? Also, how would you tell if the noise was mostly due to the magnetic or electrical portion of the RF?
Eng-Tips is the largest engineering community on the Internet
Intelligent Work Forums for Engineering Professionals
-
Congratulations waross on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!
probing for noise
- Thread starter BobM3
- Start date
ARA Technologies and other companies make E and H field probes.
I have a copy of an article from conformity magazine, Nov 2002 on "Field Probes As EMI Diagnostic Tools" - looks like it's available on the web: This shows how to make simple E and H field probes.
Another thing I do for a H-field probe is to take a short piece of coax/semirigid with a connector at one end, and solder a small SMT inductor with a value of 10 to 27 nH on the end to make quick relative (i.e. rough) measurements.
Noise can be created from magnetic or voltage sources. ICs which pull current in pulses setup magnetic fields in supply traces and low impedance output traces. High impedance inputs to ICs can be sensitive to E-field. Analyzing a troublesome circuit can require experience and can in some cases almost be a 'black art'.
I have a copy of an article from conformity magazine, Nov 2002 on "Field Probes As EMI Diagnostic Tools" - looks like it's available on the web: This shows how to make simple E and H field probes.
Another thing I do for a H-field probe is to take a short piece of coax/semirigid with a connector at one end, and solder a small SMT inductor with a value of 10 to 27 nH on the end to make quick relative (i.e. rough) measurements.
Noise can be created from magnetic or voltage sources. ICs which pull current in pulses setup magnetic fields in supply traces and low impedance output traces. High impedance inputs to ICs can be sensitive to E-field. Analyzing a troublesome circuit can require experience and can in some cases almost be a 'black art'.
- Thread starter
- #3
pstuckey
Electrical
- Oct 6, 2005
- 160
Bob,
Also bear in mind if you are trying to pick up faint signals that spectrum analyzers typically have terrible noise figures. Spec ans often have noise figures ~35 dB. So an LNA may be in order, depending on the level of signal you are looking for.
Peter
Also bear in mind if you are trying to pick up faint signals that spectrum analyzers typically have terrible noise figures. Spec ans often have noise figures ~35 dB. So an LNA may be in order, depending on the level of signal you are looking for.
Peter
Depending on the frequency, you can't do better than an 0.5 mm Pentel pencil tip in an sma-f connector. That's good from 3-30 Ghz as an efficient probe.
Even a properly designed printed circuit antenna over that frequency range is weaker than the Pentelement (my name for it).
What frequencies are you trying to probe?
Are you looking for EMI/RFI sniffers outside an electronics box?
kch
Even a properly designed printed circuit antenna over that frequency range is weaker than the Pentelement (my name for it).
What frequencies are you trying to probe?
Are you looking for EMI/RFI sniffers outside an electronics box?
kch
- Thread starter
- #6
Looking at frequencies less than 1 GHz, inside a metal box. I'm hoping to probe inside the box to look for the transmitters and then work on minimizing their effect on other parts of the circuit. Looks like the probes that Comcokid suggested will work for me.
BobM3,
In many microwave circuits, adding a cover creates EMI problems. Current moves on the cover and increases coupling greatly. Hence it's common to add microwave absorber to the cover.
I once did a coupling of two cables having gpo loads on the end. Touching the loads to each other had no measureable coupling, i.e. less than 100 dB S21 measured. Placing a box around this cable/load pair increase the S21 coupling to 50 dB. I was a bit shocked how badly an enclosure can increase coupling. This was a wideband 2-18 ghz measurement.
Is this setup of yours a set of boxes inside a larger box (like a 19" rack)? or is this a printed circuit single item inside one box.
It may be difficult to find the culprit.
kch
In many microwave circuits, adding a cover creates EMI problems. Current moves on the cover and increases coupling greatly. Hence it's common to add microwave absorber to the cover.
I once did a coupling of two cables having gpo loads on the end. Touching the loads to each other had no measureable coupling, i.e. less than 100 dB S21 measured. Placing a box around this cable/load pair increase the S21 coupling to 50 dB. I was a bit shocked how badly an enclosure can increase coupling. This was a wideband 2-18 ghz measurement.
Is this setup of yours a set of boxes inside a larger box (like a 19" rack)? or is this a printed circuit single item inside one box.
It may be difficult to find the culprit.
kch
- Thread starter
- #8
Yikes, this RF stuff is hard! I bought some probes and I hope that I can start investigating next week. My setup is a pair of boards inside an aluminum box. I need to keep the noise generated by one of the boards from influencing the circuits on the other board. I'm going to look into absorbers as you suggested.
Why is this even a mechanical problem? Shouldn't your EE be working this?
Additionally, if your goal is to prevent interference, measuring RF is a secondary problem and possibly irrelevant.
First, you need to simply jam both boards together and see if they interfere. If they don't, then you're done. If they do, send them back to the EE.
Your RF measurement is an isolated measurement with no solid referent. Let's say you measure a spike at 10 MHz. Is that a problem? And how much a problem? Can you trace it to a specific failure mode?
TTFN
Additionally, if your goal is to prevent interference, measuring RF is a secondary problem and possibly irrelevant.
First, you need to simply jam both boards together and see if they interfere. If they don't, then you're done. If they do, send them back to the EE.
Your RF measurement is an isolated measurement with no solid referent. Let's say you measure a spike at 10 MHz. Is that a problem? And how much a problem? Can you trace it to a specific failure mode?
TTFN
Before we get too deep into the RF world, lets go back to what kind of "sensitive circuit" issue you might be working with. Do you know for sure the issue is RF?
Several times I've encountered circuits that were sensitive to static charges on their plastic enclosure, or circuits that had open inputs on logic that could be triggered by nearby noise. Sometimes the noise can be very low in frequency. One time it was the magnetic field from a power supply 60 Hz transformer that when more current was pulled, the transient field coupled into a piece of equipment in the same rack causing a logic fault. Sensitive circuits can take lots of forms.
On RF circuits that deal with low levels or have large dynamic range issues of extremely low level noise can be important. However, I assume that is not the issue here or this discussion would be with a RF/electrical and not a mechanical engineer. db is logarithmic. Issues of noise at -10 dBm can be solved by the "unfamiliar". Issues dealing with noise at -100 dBm or below are solved by experienced RF engineers.
Several times I've encountered circuits that were sensitive to static charges on their plastic enclosure, or circuits that had open inputs on logic that could be triggered by nearby noise. Sometimes the noise can be very low in frequency. One time it was the magnetic field from a power supply 60 Hz transformer that when more current was pulled, the transient field coupled into a piece of equipment in the same rack causing a logic fault. Sensitive circuits can take lots of forms.
On RF circuits that deal with low levels or have large dynamic range issues of extremely low level noise can be important. However, I assume that is not the issue here or this discussion would be with a RF/electrical and not a mechanical engineer. db is logarithmic. Issues of noise at -10 dBm can be solved by the "unfamiliar". Issues dealing with noise at -100 dBm or below are solved by experienced RF engineers.
If you are new to RF and need EMI/EMC testing then I suggest you go to a lab.
To prevent radiated iterference you will probably have to meet one of the many EMC specifications and a lab test is a necessity. Likewise if you need to ensure your box is safe from external interference you once again need lab testing as the equipment needed to do Conducted Susceptability and Radiated Susceptability tests is very expensive.
To prevent radiated iterference you will probably have to meet one of the many EMC specifications and a lab test is a necessity. Likewise if you need to ensure your box is safe from external interference you once again need lab testing as the equipment needed to do Conducted Susceptability and Radiated Susceptability tests is very expensive.
Hello,
Our company presently works on a fiber optic probe that will capture/determine all EMI generate by electronic board and components. The product is at a R&D stage but we should see the first production by the Spring.
Stay tune.
Pierre-Philippe
Opsens
The fiber idea sounds good for sniffing out sources. I suppose it can pinpoint the radiation location better than a normal antenna due to it's small size.
I would think there may be a sensitivity issue on how efficient your fiber probe is. Being an antenna engineer, there's some laws of physics, but being very close to a source lets you be very inefficient and still pick up the signal.
If you have the ability to receive pure linear polarization on your fiber probe, and it's efficient, near field probe measurement facilities could use the product. Also, I would hope you can have a long length of fiber before your detector for field probe use, i.e. keeping the wires away.
Sorry for the divergence, fiber probes are interesting.
kch
I would think there may be a sensitivity issue on how efficient your fiber probe is. Being an antenna engineer, there's some laws of physics, but being very close to a source lets you be very inefficient and still pick up the signal.
If you have the ability to receive pure linear polarization on your fiber probe, and it's efficient, near field probe measurement facilities could use the product. Also, I would hope you can have a long length of fiber before your detector for field probe use, i.e. keeping the wires away.
Sorry for the divergence, fiber probes are interesting.
kch
I have some very recent experience with probing some high power HF airborne equipment for RF leaks.
I made a little multi-turn loop (about ten turns of small gauge wire on a ~2-cm circle) as a probe connected to an oscilloscope. The short leads to the female BNC connector were twisted. The loop was wrapped up with flat lacing string to make it secure. I used a 10-ft BNC-BNC lead to the oscilloscope.
Although the absolute readings were variable and relative; the loop could repeatedly pin-point leaks to better than 1-cm. I was very impressed with how location-sensitive the simple loop was. It was as location-accurate as a magnifying glass. The location of the RF leakage peaks corresponded perfectly to the physical or mechanical features of the shielded equipment.
Highly recommend that you check different frequencies all over the band (if applicable). Things changed dramtically with frequency.
I made a little multi-turn loop (about ten turns of small gauge wire on a ~2-cm circle) as a probe connected to an oscilloscope. The short leads to the female BNC connector were twisted. The loop was wrapped up with flat lacing string to make it secure. I used a 10-ft BNC-BNC lead to the oscilloscope.
Although the absolute readings were variable and relative; the loop could repeatedly pin-point leaks to better than 1-cm. I was very impressed with how location-sensitive the simple loop was. It was as location-accurate as a magnifying glass. The location of the RF leakage peaks corresponded perfectly to the physical or mechanical features of the shielded equipment.
Highly recommend that you check different frequencies all over the band (if applicable). Things changed dramtically with frequency.
- Status
Similar threads
- Locked
- Question
- Locked
- Question
- Locked
- Question