EMI Noise Reduction Design Ideas for PI Detectors

[Elliot]

Along my journey to create the best PI machine using regular electronic components (i.e. no ADC's or microprocessors etc.), I have reached a point where my limiting factor to better sensitivity (SNR) is the need to eliminate/reduce EMI that is picked up by the search coil.

Does anyone have any useful and practical design ideas/tips that I can look at please?

[green]

Along my journey to create the best PI machine using regular electronic components (i.e. no ADC's or microprocessors etc.), I have reached a point where my limiting factor to better sensitivity (SNR) is the need to eliminate/reduce EMI that is picked up by the search coil.

Does anyone have any useful and practical design ideas/tips that I can look at please?

Figure8 Rx coil reduces EMI pickup but it has it's set of problems. Would be interested if there is a way to shield EMI.

[waltr]

Along my journey to create the best PI machine using regular electronic components (i.e. no ADC's or microprocessors etc.), I have reached a point where my limiting factor to better sensitivity (SNR) is the need to eliminate/reduce EMI that is picked up by the search coil.

Does anyone have any useful and practical design ideas/tips that I can look at please?

What is the EMI source?
Is AC Mains (50 or 60Hz) then having the Sampling rate a multiple of the EMI rate helps.
This is why some detectors (TDI for one) have a pulse rate adjustment pot.

[green]

What is the EMI source?
Is AC Mains (50 or 60Hz) then having the Sampling rate a multiple of the EMI rate helps.
This is why some detectors (TDI for one) have a pulse rate adjustment pot.

What does integrator out look like if AC Mains is the source of the noise? Does the noise repeat or is it random. I tried adjusting sample rate, didn't see a reduction in noise. Maybe adjustment was too course. Some scope pictures. An example of what I see. Need digital scope or CRT with persistence to see 5sec trace. Think need 5 or 10 sec trace to see what is happening. Need to look at integrator out. Some thoughts maybe correct maybe not.

Scope and lamp probably not the only noise source.

[waltr]

If just AC Mains then there will be a 'beat' frequency that is the Sum & Difference of the pulse and AC frequencies.
As the Pulse rate is adjusted the beat frequency decreases until it becomes DC- no interference.

Your scope shots are shows 'random' type noise. CFL lamps typically have a switching power supply that is not frequency controlled and can have lots of wide band noise. Nothing you can do to keep this out of a detector except lots of distance since a PI detector is basically wide-band.

ONLY single frequency (AC mains, etc) can be nulled with pulse rate adjustment. Wide band Noise can not.

FYI: There are many CFL lamps sold that do NOT pass FCC or CE emissions standards. These are actually illegal to be sold but are sold in many places. Ham Radio operators commonly find these 'bad' CFL lamps disrupt radio communications. Only solution is throw them away.

[green]

Thanks, Do you have a scope trace of integrator out showing mains pickup and no mains pickup. My bench circuit with integrator isn't working. Need to get one working to test the bigger coils I'm making.

[waltr]

Thanks, Do you have a scope trace of integrator out showing mains pickup and no mains pickup. My bench circuit with integrator isn't working. Need to get one working to test the bigger coils I'm making.

No I don't have a scope pic since there is much other noise in my workshop and is it is an old analog scope that does not do well with very slow scan rates. The Beat would look like a slow Sine wave if ONLY AC mains pick up.
If it picks up broad band noise then might not see it.

I once tried detecting a local pack that has high tension lines over head. The audio out of my HH2 detector was a few Hz modulation.
I didn't want to open the box there and try tweaking the pulse rate pot (it is not panel mounted like the TDI).
This is the only time I actually heard AC Mains beat frequency interference.

In the shop you can probably induce AC mains interference with an AC Mains powered inductor. Maybe an AC motor would do this.
Then move the detector's coil closer while watching the O'scope.

[Vadim18]

What if put small EMI pic-up coil (with same bandwidth as main coil) to the shaft, and than tune final EMI supression level with two potentiometers - phase shift pot and amplification pot. Of course this is not proven idea, just part of my fantasy.

[green]

Has anyone done testing to see what frequency band typically causes the most noise signal at integrator out. 1 to 10kHz, 10 to 100kHz or 100 to 1000kHz?

[Elliot]

What is the EMI source?
Is AC Mains (50 or 60Hz) then having the Sampling rate a multiple of the EMI rate helps.
This is why some detectors (TDI for one) have a pulse rate adjustment pot.

I have this in my design too (+/- 48 hertz adjustment - covers both 50 and 60Hz AC at some point of adjustment), but to get the best out of it, you'd need to sync the AC phase and pulse cycles - much more tricky! My biggest issue is noise around 1.1 -1.2 Mhz. I can see this low level noise in the coil flyback response. Who knows where its coming from though?

[green]

I have this in my design too (+/- 48 hertz adjustment - covers both 50 and 60Hz AC at some point of adjustment), but to get the best out of it, you'd need to sync the AC phase and pulse cycles - much more tricky! My biggest issue is noise around 1.1 -1.2 Mhz. I can see this low level noise in the coil flyback response. Who knows where its coming from though?

Does integrator out noise level change if Tx is disabled? Maybe remove Tx MOSFET or what ever works with your circuit.

[Elliot]

There is no doubt that a Tx fine frequency adjustment pot works to pretty well reduce mains 50/60Hz noise, but has anyone got a good solution to reduce random noise also? How does a Minelab Equinox do noise reduction (not a pi machine, I know)?

[green]

There is no doubt that a Tx fine frequency adjustment pot works to pretty well reduce mains 50/60Hz noise, but has anyone got a good solution to reduce random noise also? How does a Minelab Equinox do noise reduction (not a pi machine, I know)?

A figure8 Rx coil is the best way I've found to reduce(not eliminate)random noise.

[Carl-NC]

There is no doubt that a Tx fine frequency adjustment pot works to pretty well reduce mains 50/60Hz noise, but has anyone got a good solution to reduce random noise also? How does a Minelab Equinox do noise reduction (not a pi machine, I know)?

Equinox does a fine frequency adjust. Random (non-periodic) noise is tougher to deal with, especially in a wideband system where filtering options are limited.

[green]

Equinox does a fine frequency adjust. Random (non-periodic) noise is tougher to deal with, especially in a wideband system where filtering options are limited.

What are the filtering options for a PI?