attaching probes for high frequency measurements 3

[BobM2]

I need to investigate some noise problems on a printed circuit board. I'm adding an RF receiver to a device that has a commutated motor - lots of high frequency noise. Any suggestions for the type of probe and the way to attach to different areas on the PCB? I'm interested in noise to 500 Mhz.

 

[logbook]

It is difficult to say what signal levels might be a problem. Normally the problem is common-mode noise in the system since the test equipment will be earthed. For broad band noise(but not down to DC) I use a small RF transformer stuck onto a piece of pcb and feeding coax into the scope or spectrum analyser. The RF transformer gives the single-ended scope/analyser a differential input. However this type of assembly might be a bit ‘technical’ for a mechanical engineer.

You should start off using a 10:1 scope probe, but throw away the earth clip, dismantyle the probe and connect to the inner grounding ring. If that is still too noisy then go for a straight coax lead soldered to the pcb, with say 470R in series with the signal wire at the signal source end. This again makes a 10:1 (-ish) probe but with better common-mode performance.

After that you are going to have to buy a differential probe, but now you are probably talking about £1000+

This is not a trivial measurement problem.

 

[Comcokid]

For quick probing, I use a "sniffer" made from a 0603 or 0805 size 10uH inductor attached directly to the end of a piece of small diameter semirigid coax - the center conductor extended just enough to attach to one end and the other end of the inductor soldered directly to the shield. Connect this to a spectrum analyzer to look for the offensive signal. You do not have to make direct electrical contact to inductors and traces to "sniff" the signal at a point. This setup is mostly sensitive to the magnetic field at different parts of a circuit.

Oscilloscopes may not have enough dynamic range to isolate a interference problem, and scope probes being sensitive to the electric field may pickup too much noise to effectively isolate some problems.

 

[BobM2]

Wow, I didn't realize the probe would be so costly. I think I will try the ideas the both of you suggested. I assume the difficulty with the RF transformer is getting the frequency response flat beyond my frequency of interest and also getting the input and output impedances correct.

Thanks to the both of you.

 

[logbook]

Any half-way decent modern digital storage oscilloscope will have an FFT capability, so you will be able to get sub-lsb resolution by using a 4096 point FFT. The FFT resolution will probably be better than 0.1 LSB, but then there is the question of the noise level of the acquired data.

 

[BrianR]

If the signal is big enough to be seen on an oscilloscope it is certainly strong enough to really worry your receiver. The small inductive sniffer probes described attached to a spectrum analyser are what you need. The frequency response of the probe is not too important for a first level check after all you are not taking quantitative measurements at that time. I have used a length of thin coax with a 1" to 2" dia loop made by shorting the centre conductor to the braid for tests like this but never tried the mini inductor which sounds like a very good idea.

 

[BobM2]

Thanks to all of you. I'm an electrical wannabe so this stuff doesn't come easy to me. HP's probe section also had technical articles on sniffers. My question on the use of a sniffer would be: How do you know the magnetic field you are seeing is from the conductor you are interested in? Couldn't it be from another conductor elsewhere? Would I have to move the sniffer around and rotate the coil to try to determine where the field was coming from? Too hard for a mechanical - I need pictures!

 

[logbook]

You are right to think you have to move the sniffer. The field will decay with distance from the source. As you get closer the field gets bigger. This enables you to quickly find the offending component or tracks. Now if your sniffer is 8 feet in diameter, you will not be able to get positional accuracy of 1mm! This is why a small inductor is a good sniffing device.

People do this type of seeking all the time with metal (treasure) detectors, so I don’t really think you need a picture

 

[csharp]

Hello,

If you decide to use passive probes a
very handy thing to have around is
the Bendix connector. You can buy them
from Specialty Connector Company or from
Tektronix. It's a small, bulkhead-
mountable, metal socket that the probe
tip fits into and it's terminated with
a solder post. You can't get a more
secure or shorter connections to the probe.
I usually solder the body of the connector
directly to the ground plane.
I've used these with both passive and FET
probes on sub-nanosecond logic as well as
for RF measurements and they work extremely
well. They're also great for taking
measurements in electrically noisy environments.
It beats the heck out of using a coiled
up paper-clip.

Best regards,
csharp

 

[toulouse]

if you want some explanations, look at

http://www.emcesd.com/pdf/iprobe98.pdf

, and in general at

http://www.emcesd.com

Thomas K.

http://www.emcrf.eng.br

ICQ#282786426