noise sources in the PI – front end... what is your experience?

[wm]

Welcome back Eric and hi everybody!

My question refers to the possible noise sources, which may appear in a OPA-FE when designing a high sensitivity detector.

Also a special question refer ed to the search head – since there is a audible sound when powered – could the vibration of the wires in the search head cause a performance degradation?

cheers
Walter

 

[Eric Foster]

Hi Walter,

Most of the noise is electromagnetic induction in the search coil of radio transmissions, power line noise and all manner of other electrical and electronic devices. Next in line is noise generated with the detector, and assuming all the voltage rails are noise free, then the front end amplifier and its associated resistors need careful selection. Even the transmitter circuit, although in the off state when receiving, is another path for noise to enter the receiver.

Regarding the vibration of the coil wires, this is due to the reaction of the pulsed magnetic field from the coil and the earth's magnetic field. The higher the pulse current the greater the buzz, hum, or whistle that you get, depending on the TX frequency. Upwards of 20A peak current can make quite a noise. However, when potted, a coil is usually quiet. I have not noticed any performance change caused by this.

Eric.

 

[wm]

Hi Eric,

thank you for taking time to answer my question!
In the (long!) meantime I was trying to recheck my circuits and precise my understanding.
Leading assumption for my design is the possibility to use RF-noise cancelling coils and bipolar pulse excitation (which cancels static and low frequency fields).

Similar to your Beachscan design, I am aiming for high pulse rate (5-10kHz) rather then high currents. This is expected to further lower the interference noise level, rejecting frequencies below the pulse rate (believed to contain most of the disturbing power – does this match to your experience?).
That is where the noise, generated by the electronics itself starts to become significant.
I found so far lots of information on improving the FE-amplifier, but the noise contribution of the sampling circuit and the subsequent integrator seem underestimated.

Considering that sampling is done only a small fraction of the time (say 20us in a 200us period or 10%), the off-leakage of the switches is active all the time, pumping noise into the integrator 90% of the period time. The SD2000 design uses standard 4066-switches with specified leakage currents up to (max value) 1uA. Considering the signal levels at that point, comparable low frequency leakage noise may become killing point. Unfortunately, this is not specified.
The GoldscanIV design uses J113-jFets, which (in my understanding) should be superior in terms of noise. Is this the reson for using jFets?
Did you (or somebody else) investigate this issue? Your comments would be very helpful.

Walter.