Probably this?
These data aren't directly applicable to loop antennas, because the winding doesn't usually cover the whole core; the A_L is higher in the center than at the ends.
Tim
-- Seven Transistor Labs Electrical Engineering Consultation Website: http://seventransistorlabs.com
Spoon waving across the lab?! That's impressive. I would have thought you'd have said, "...a screwdriver waving across the lab." But what you have in your lab is your business :)
Have you any need to 'calibrate' the sensitivity?
Two ways to go:
the 1st method is by far the easiest and will get you maybe 3-10% accuracy, but perhaps as close as you want/need.
The 2nd one, you can get to within 0.1% but requires a bit of care.
If you want to do the Helmholtz coil, the easiest would be the simple construction of two coils. Keep in mind that the volume that is within the 'constant' field is less than 1/3 the dimension of the outside. For example, a 6 foot diameter, 6 foot tall pair of coils has less than a 2 foot cube of really good field space inside it. But at least you can calculate everything and accuracy is a function of construction accuracy and measuring the freq and current of the driver.
I mention Helmholtz-type because there is a type of coil that has 'distributed' windings down the cylindrical dimension that creates a larger volume of accurate field, but that requires more finesse to construct.
You can do an 'estimate' of a coil with a core: Find N, permeability, core cross section, core length to diameter ratio. Now assume that the core distorts space by the amount of its permeability times the square root of the core length to diameter ratio and take that times the core cross section area. THAT is roughly approximately the equivalent area *IF* the core were air core. You have N, so you can now 'estimate' the equivalent sensitivity.
When I say, estimate, I mean estimate. The above equation was empirically derived from very few observations, not based upon any rigorous relationship, nor upon curve fitting to many, many equivalents. But does take into account [well, kind of] what permeability means. And when the equation got me to within 5% accuracy, I kept it around as 'interesting'.
Apologies to jump in here, but why only dessimate 1 in 20? why not 1 in
40, or 1 in 160?On my soundcard, I'm running often at 91%, usually 89% of Nyquist with no problems, so getting close to Nyquist limit should not pose a problem. At
1 of 160, you'd have approx 275 samples per second, with the Nyquist limit near 140Hz - the max interest is 40Hz, plus you're trying to filter all above 40Hz anyway.That rate, for a 16 bit digitizer, translates to around 550 bytes per second, really slow rate.
Really close. From memory, last page in Fairite manual, back in hard copy days.
The plot took up the whole page with lots of curves asymptotically approaching 'no effect'
On Wednesday, October 30, 2013 10:27:33 PM UTC-4, snipped-for-privacy@comprodex.com wrot e:
Fun thread (and project.) (Martin raises this same question below.) Why the need for 'real time'? I 'm assuming someone remote will want to try and correlate your measurement, with something they are seeing locally. The simplest answer to this would seem to be the time stamping that Tim W. suggested above. Is there some r eason that doesn't work?
Lurking, George H.
At those frequencies, I'd use an oversampling A/D and let the internal decimation take care of the initial filtering. It is not difficult to make a good pre-sampling filter for such A/D so that it does not have any measurable effect to the ELF phase.
It is far simpler to post-filter with a linear-phase FIR brick wall after digitizing, as well as decimate the signal as needed. The same applies to notching the power line and its harmonics.
The filtering will always cause some delay to the signal, unless the OP has a time machine to guess the samples to come next.
I just wonder what the 'complex waveform' of the OP is. For any non-sinusoidal repetitive signal it means harmonics. The question is: how many of them?
-- Tauno Voipio
It isn't clear from the context now whether the OP means the signal is mathematically complex as in Z = (x + iy) - eg phased data out of an interferometer baseline correlator or complex as in aperiodic noisy waveform having a randomly varying mixture of frequency components.
I had assumed the latter - a random real valued variable time series containing significant frequencies above the filters nominal cutoff.
The bit I really don't understand is that most of the interesting dispersion effects of ELF whistlers in are in the 10kHz to 1kHz decade. eg
see fig 1 on p40
I don't understand why the filter cutoff at is set at 40Hz.
-- Regards, Martin Brown
We sold all the screwdrivers when we stopped using wood.
Thank you for the info. No calibration is required since this is not a measurement system. We are only interested in the frequency components of the signal.
Mark Harris
Without knowing the exact requirements, if there is phase shift at some frequencies, would it be enough that the phase shift is well known for a particular frequency across _all_ coils (channels) ?
If this is the case, do some low pass filtering in the analog domain well above the intended final cut-off frequency. After oversampling ADC, filter to final bandwidth with a predictable phase shift on each channel.
I have noticed AC coupled input devices seem to accept a mix of ELF's better than a single ELF alone. I attributed this to the added complexity, or density, of the waveform.
For example, if you mix enough ELF's together, you can play the synthesized signal through a standard audio system.
The would be fine for our application. I am relieved to hear our goal can be achieved with existing methods.
Mark Harris
I believe I did mention it was an EEG-like "complex" signal.
To filter out the power grid, and there are no significant Schumann, Earth ionopshere cavity, resonances above 40Hz.
Mark Harris
A number of respondents have asked this. The project involves monitoring of biological functions in lab animals in response to instantaneous geomagnetic activity. The subjects are distributed around the world, hence the need for streamed data.
I am not proficient in DSP or FIR filter design (did anyone notice?), so I was leaning toward the least exotic approach.
What I really need at this stage is block diagram representation of the different options for the entire system, so it can be established where additional expertise may be needed.
We already have the XY air coils, our exisitng EEG-type amplifer, and a DC modified soundcard (or commercial DAC module).
Leaving the compression issue aside for the moment, or whether or not it is necessary, that leaves the minimal phase distortion filtering. There have been so many good suggestions made here and I now need to choose the one that most efficiently serves our purpose.
Mark Harris
Being an Analog Designer, I highly recommend moving all this into the digital domain as 'quickly' as possible.
You have so much power from using that soundcard for data acq. I'd go straight into it to start with! Any potential code you'd write [maening FIR, exotic filters, etc] you can completely simulate and verify using free Matlab clone, octave. Even play with representative waveforms and create representative results. Just 'walk before you run' for using software AFTER the soundcard you can keep updating the system, tweaking its performance. Difficult to do with soldered hardware. You could end up spending more time maintaining electronics, then doing what you want.
After creating results, no idea how to put that information into internet packets, but sounds like people here have that solution ready for you.
You do know that the earth's magnetic field varies around the globe? Like from less than 25 to more than 100uT? And of course the magnitude and the phasing here does NOT mean the same magnitude and phasing elsewhere. Raising the question, IF you want to relate field to animal subjects' actions; shouldn't the measurement of the field AND the subject be in the same location?
I think you ought to look more closely at why this is... to understand just what effect it is that you are actually observing.
An issue here is the meaning of the word "mix" - it has two very different meanings, depending on how you use it.
In the usual audio terms, "mix" refers to a purely linear process - you are just adding the two signals together (their amplitudes add, at any given instant). This process does not create any new frequencies... the "mixed" signal contains all of the frequencies which were present in any of the inputs, and none other.
If you were to take these input signals, carefully low-pass filter them down at 40 Hz, and then "mix" them by adding them cleanly... then the resulting composite signal will have no frequency energy above 40 Hz.
The other form of "mix" - the one which is often used in radio technology, as in an "mixer" stage in a receiver - is a highly nonlinear process. It's not addition, but multiplication (or an operation much like it). This type of mixing *does* create an output which contains frequencies that are not present in any of the inputs... in general, for any two frequencies F1 and F2 that were present in the input, this type of mixer will give you an output with F1, F2, F1+F2, and F1-F2.
So, if you "mix" two ELF signals together through this sort of process (e.g. through anything which adds nonlinear distortion), you will create new mixing-product frequencies which are higher than those present in the inputs. For example, if one ELF had some 30 Hz content, and another had some 35 Hz content, a nonlinear mixing will create a small amount of 65 Hz signal.
It is possible that this is what you are hearing, when you say that you can "play the synthesized signal through a standard audio system". If so, all you are hearing is distortion - you're hearing something which wasn't present in any of your actual inputs, but was created by the mixing system itself.
Some AC-coupled devices use a poor AC-coupling technique: they just use an electrolytic capacitor, with no DC bias voltage. Such capacitors can create some non-linear distortion when the voltage across them is of the wrong polarity.
Overdriving an audio input to the point where it starts to distort can have the same effect.
A clean, low-distortion, linear "add the signals together" signal processing chain won't do this to you.
You *still* haven't defined what *you* mean by "complex"!
I suspect you just mean a simple boring real valued time series of measurements with some sort of roughly inverse power law distribution of power with frequency and some interesting features here and there.
Feed the signal into something like Daquarta in its realtime FFT spectrum analyser mode and then you might see something interesting.
Precision instruments often measure signals synchronously with the local mains power so that to first order the mains fundamental and all its harmonics cancel out. Solarton 7060 DVMs did this ages ago. This would fix your sampling rate at 30Hz in the USA and 25Hz in UK and a corresponding Nyquist frequency of half that which may be too slow.
-- Regards, Martin Brown
Only if by chance the phases happen to be right to make the synthesised waveform spiky. eg
sum n= 1..inf Cos(nt) = Kronecker_Delta(t)
Is a very sharp spike at the origin.
It is a fundamental tenet of all linear systems and of Fourier analysis in particular that you can add phased sine components in an orthogonal basis function together and they remain completely independent.
If you are seeing something else then there is something wrong with your setup - probably intermodulation distortion or clipping generating harmonics of the ELF that are in the audio band.
Two similar frequencies will appear to generate audible beat amplitude modulation as there difference frequency but they still decompose precisely to the original sine waves in Fourier analysis.
The identity
Sin(A) + Sin(B) = 2Sin((A+B)/2)Cos((A-B)/2)
explains why this occurs
Put a spectrum analyser onto the thing and see what is going on!
-- Regards, Martin Brown
Are the sensors in the same place as the animals ?
If so, the animals are also subjected to the mains generated magnetic field. Why do you want to filter out the mains generated 50/150 Hz or
60/180 Hz magnetic fields. Any biological effect would be the sum of the geomagnetic and mains generated magnetic field.Especially, if the mains generated field is much stronger than the magnetic field, wouldn't that destroy any validity of the results ?
Thus the tests should be conducted in a place with low mains noise, so that neither the animals nor your instruments should be affected by the mains.
Why would you expect the behaviour or biological functions of lab animals scattered around the world to be related to small local ELF fluctations measured in some other random location?
They are far more likely to be affected by the nearest aircon compressor motor or passing electric cars, trains and trams outside. They have problems with passing electric trains at Jodrell Bank.
This is almost as wacky as monitoring catfish in the hope that they can detect earthquakes coming. Although my own personal experience is that sometimes you can hear an earthquake just before the shockwave hits. I was in Japan when the world moved abruptly 1" to the west and I swear that I heard it coming although I didn't know what it was. Only ever happened once - all the other earthquake some quite big were normal.
Simple analogue Bessel filter to get bandwidth down to ~10kHz digitise. Examine what signal you have with a spectrum analyser.
If the rig is sensitive enough you should see whistlers which will be a good test of the systems sensitivity.
Then decide how hard you need to try to get rid of the mains hum.
You might also want to consider having a pre-emphasis gain that varies with frequency to make the best use of your DAC.
-- Regards, Martin Brown
I confirm the sensation feels like hearing, but not real time, more like 'what was that?' Another sensation is feeling the ground move abruptly, like having it ripped up/down or side/side just like a small roller coaster ride. You have no idea how intense the cues were before Japan's big quake! That was like screaming silence.
After having lived in CA for 35 years, got used to 'sensing' when a quake would come, could pretty much tell how big, when, and to some degree where, so we'd go around the house putting breakables down to the floor to save them from scuttling off a table, etc. You know it's going to be close when you see a lot of spiders out and about, too.
Worst part is the sensation expanded to now include global, we watch the news to find out WHERE they hit now so are starting to sort out what what means for distance/location.
Oh Dear, A big piece of iron moves in NYC and rats in Tokyo crap more often? ("Are you going to account for speed of light effects", he asks sarcastically)
George H.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.