EMI-free test chamber?

[Carl-NC]

Still trying to understand why. Thinking sampling theory. Wondering what causes beat frequency. Two frequency's summed causes beat frequency, something else? Wondered if aliasing could be causing it. Excel program calculates frequency out with 60Hz and odd harmonics. Don't see that being the reason. What is causing the beat frequency? Does the sample frequency need to be an exact multiple or is .1% close enough?

Does the sample frequency being in sync with the 60Hz reduce he noise, cancel the beat or both?

Yes, it's due to aliasing. If exactly synchronized to 60Hz then it becomes a DC offset. If very slightly off, then the beat frequency can be low enough that the SAT circuit removes it.

[6666]

A good way to eliminate 60Hz interference in a PI detector is having the Period an exact multiple of 60Hz period. 666.66... us

Yep its a good frequency

[waltr]

A good way to eliminate 60Hz interference in a PI detector is having the Period an exact multiple of 60Hz period. 666.66... us

Yep its a good frequency

lol

[green]

Scope trace(60Hz_3)with soldering iron by Rx coil. Tx coil disconnected, 60Hz+noise.

spice simulation(60hz)60Hz sampled at 1000 and 1499.925 samples/second. Don't see a difference other than 60Hz signal is higher with 1499.925 sample rate due to higher integrator gain.

spice simulation(60Hz_2)60Hz+noise sampled at 1000 and 1499.925 samples/second. Don't see a difference other than 60Hz signal is higher with 1499.925 sample rate due to higher integrator gain and sample rate effected noise signal.

Yes, it's due to aliasing. If exactly synchronized to 60Hz then it becomes a DC offset. If very slightly off, then the beat frequency can be low enough that the SAT circuit removes it.If chart(aliasing_1) is correct the only odd harmonic that becomes DC offset is the 25th.

Don't know if simulation would be the same as detector. Don't see a difference but maybe can be heard? https://www.geotech1.com/forums/show...116#post239116 maybe waltr could show a scope trace of amplifier out and integrator out with sample rate in and out of sync.

I believe it can be heard, just don't see why. I would try but the only circuit working right now just has amplifier out, no integrator. Didn't have a 60Hz problem when detector was working with 1000samples/second. I wouldn't have heard it because I turn a led on when signal>threshold(bad hearing). Didn't see any beat frequency when looking at last integrator stage with a scope.

[Altra]

A good visual is to set your scope trigger to "AC Line". Put a scope probe on the TX logic line. When the TX frequency is synchronized with the mains frequency it will not drift horizontally. The more out of sync, the faster the drift, the same fluttering that you can sometimes hear in the audio. You can see the drift over time with your circuit and the power company time bases.

I would like to know how to get rid of "telegraph noise'. You can get rid of the 50/60hz pretty easy by synchronization. But the sporadic noise that hits at curtain times is discouraging.

[waltr]

spice simulation(60Hz_2)60Hz+noise sampled at 1000 and 1499.925 samples/second. Don't see a difference other than 60Hz signal is higher with 1499.925 sample rate due to higher integrator gain and sample rate effected noise signal.

In the 60Hz_2.png there is a lower frequency in the plot, V(out) blue trace, that has a much higher amplitude. This is the 'beat' or Aliasing due to PI rate not a multiple of the 60Hz.

A good visual is to set your scope trigger to "AC Line". Put a scope probe on the TX logic line. When the TX frequency is synchronized with the mains frequency it will not drift horizontally. The more out of sync, the faster the drift, the same fluttering that you can sometimes hear in the audio. You can see the drift over time with your circuit and the power company time bases.

I would like to know how to get rid of "telegraph noise'. You can get rid of the 50/60hz pretty easy by synchronization. But the sporadic noise that hits at curtain times is discouraging.

Yep, 'flutter' is a good term for what is heard.

What is "telegraph noise'?

I do get extreme noise when I am not far from a road and a car/truck with a hot ignition system goes by. This is Wide band, high Voltage noise and not much can be done about it.
Possibly a very sharp band pass filter after the integrator at the PI cycle rate may cut this down but if it Aliases into the PI's pass-band then nothing will remove it after the detection stages.

[Altra]

What is "telegraph noise'?

Sounds like Morse code that breaks up a smooth threshold. Remove the coil it goes away.

Here is an article on noise related to semiconductors. Mentions "telegraph noise" . Maybe I am hearing a ham radio operator on hf. A second article on popcorn noise.

[waltr]

Good articles, I saved them.

It is possible you are picking up a local ham radio. Does it do this if you are several km away?
Know anyone that can read Morse?
It does sound like it is being picked up by the coil.

Another possibility is that you have a bad component or a cheaply made one. Do you buy Ebay/Chinese or good name brands?

From the articles I would try changing the coil MOSFET to a different make, part number. If now gone I'd say a bad MOSFET.
Also if NO coil then NO current through the MOSFET so worth a try.

As for other 'noise' I did try different JFET op-amps in the integrator & SAT. The newer ones did have less total noise. ei: TL082 better than TL071 even though specs are identical. Analog switches may also be different between makes, there are many 4066 switches on the market.

I do not run shielded coils but maybe shield will help or maybe shield makes it worse.
I understand how and what a coil shield does but haven't had any of the issues a shield solves.

I do have the circuits in a metal box that is connected to the circuit "ground".

Maybe also try this thread original question: put the coil inside a earthed metal enclosure (chicken wire cage should work) to eliminate coil pickup of outside interfere.

[dbanner]

What is commonly referred to as chicken wire, this will work. It's a very fine wire mesh. Think of emi in terms of wavelength, the higher the frequency, the shorter the wavelength. Wavelength in meters is the actual length of one cycle. As log as the wavelength is considerably longer than the spaces in the chicken wire, the wave cannot pass through, it is diffused(scattered). Result is substantially less interference.