ML EQUINOX coil discussion thread

[Skippy]

The divider is two 1k8 resistors, with a (probably quite big) cap to GND for decoupling, this is brought out to PCB pad 'REF' , and connects to pin 7 of the dual opamp. Pin4 of this opamp is the -ve supply, connected to GND, pin10 is the +ve rail, to 3V3. The 'power-down' pins, 5 & 6 are both tied to +3V3.
Here's the datasheet for the opamp ( if Foma is correct ):
Texas OPA 2835:
http://www.ti.com/product/OPA2835

The VSSOP package is the one used.

@Nordic: You won't see the smallest detectable signals on a scope, that's for sure. On a typical single-freq machine with plain coil, changes of 5 microvolts will be detectable. If this coil has a pre-amp gain of about 15, then maybe 50 microvolts out of the amp may be detectable, just a guess.

[nordic]

Thanks Skippy. The sound from the control box is way before TID is displayed. Changes are tiny, it was even reacting to the scope probe somewhat.
It is interesting to see that when TX circuit transconducts, coil oscillates and there is a small spike on the RX coil.

[nordic]

I played a little bit with the 6 inch coil yesterday. I've noticed that the VDC cable has a lot of ripple in it, around 60-70mV, not sure whether it is due to crosstalk or sagging, but it was contaminated with the TX coil signal. I could bring the noise down to about 10-15mV with some filtering, but oddly, all targets on control box started coming in the 11-13 range! The other thing, I tried change coil voltage to 3-4V, but on extremes it subtly reduced the air test depth, the best distance I was getting at around 3.6-3.7V, which is a battery voltage...

[dbanner]

I played a little bit with the 6 inch coil yesterday. I've noticed that the VDC cable has a lot of ripple in it, around 60-70mV, not sure whether it is due to crosstalk or sagging, but it was contaminated with the TX coil signal. I could bring the noise down to about 10-15mV with some filtering, but oddly, all targets on control box started coming in the 11-13 range! The other thing, I tried change coil voltage to 3-4V, but on extremes it subtly reduced the air test depth, the best distance I was getting at around 3.6-3.7V, which is a battery voltage...

Perhaps there already exists some filtering in the control box that takes care of that tx noise. But also consider mismatch of impedance.

[nordic]

Need someone's educated opinion. Is there any drawback nulling the RX coil using a small ferrite attached to the RX coil? Here are few screenshots of the scope before and after finding a certain place for the ferrite.

What I see with air tests, that control box is a lot more stable with TID even significantly away from the coil. Haven't done any real ground tests though. Or is this all just worthless time waste...

The other thing noticed, gold modes have extremely low background noise.

Many thanks!

Default, multi freq


Moving ferrite around to get the lowest Vpp


End result

[Skippy]

I think it's correct to state that ML have nulled the coil correctly/adequately, from the factory. Any "Improvements" you make will have no performance benefits even if they appear to give a lower signal from the coil
Adding ferrite ( and non-iron objects ) can affect the null, but you have to consider the wide freq range. It's no use improving the 8.2kHz null if the 39 kHz null gets worse, etc. If you have the 800 model, you can choose all the freq range, from 5k to 40k, to try out various nulling techniques. Owners of the 600 model are restricted in this sense.
And I have to say - that effing big lump of ferrite is not appropriate. The pieces I've used on other coils have been 6 x 6 x 2 mm, it matters where you place it, of course, but in general, the less you use the better. And I should remind you that the coil needs to be kept away from all metal objects when trying this kind of nulling. No point nulling out the legs on your workbench.

When we eventually do try and make our own coils, this 'bits of metal' nulling technique has some use.

[nordic]

Thanks Skippy, much appreciated input. I don't call it an improvement, as I don't know how it will perform, but will give it a go. Kinda like this empirical stuff. I have no doubt ML squeezed most if not all from the setup.

I haven't noticed any drop in distance during air tests, sure that is not on real ground, so time will tell. What I have noticed, that the ID would come sooner as soon as I hear iffy sound, it gets the TID rather quickly and close to real value. Without nulling, TID would jump around until object gets close enough. Gold modes on 800 were quite unusable to me due to background noise, but it is now very faint, that is actually nice. It looks like Gold modes feed audio directly to ears.
I've tried all frequencies, all of them seem to have dropped significantly, only 40k has the largest Vpp out of all, but it is multiple times smaller than before. As you have suggested, I have tried smaller ferrite pieces I had, but they could not bring Vpp low enough, perhaps it's just the distance from the coil.

[EL NINO]

My experience with the multi-frequency Spectra V3 and the 13 "ultimate coil balance - with the help of the Signal Probe function - implemented in Spectra - showed that the coil, which is heavily balanced with a few 50 eurocent coins, has not lost sensitivity or reach in field tests. range was quite comparable to another 13 "coil, and better than a 12x10 sef coil ..



The initial value of the coil balance in the signal probe was above 65% of the signal at RX gain 9 ...
After adjustment, the coil balance has improved by more than 2X. The coil could then work on the maximum Rx Gain 15 ..




Field tests have shown the correct sensitivity even the depth of the 13 "coil ultimate ..






I would like to say that the work of balancing the coil at the multi-frequency has its meaning .. ,, and I think it will mainly manifest in more mineralized terrain, or in the separation of rough iron ..- where it is more resistant to overload ,, ...So each coil when working above its signal-before overload .. will identify non-ferrous objects as iron...

[nordic]

Thanks EL NINO. Did you notice any improvement in MD operation? I did not really. However if I make nulling twice as bad using same technique, there is a drop in detection depth.

There is a noticeable change to a positive side if to tackle the noise in the Vdc rail, because by default the line appears to be quite noisy. It is also not exactly 3.3V, wonder if it is regulated sufficiently. With all the gain in the coil and the digital domain, that can't be good. I would expect the ADC analog section to be fed with the same line and it does not add anything good to SNR figures.

After applying some simple filtering techniques, I have noticed increased sensitivity to small things in front of the coil. Before, it would simply not react to them. In the field testing, the iron became identifiable clearer somehow, and generally MD was running quieter on max 25 sens.

Just a pic of it covered from the rain, sitting between the coil and the unit.


If to look at the TX coil firing timing, there is significant jitter present, but I'm not sure if it is a problem or not. My assumption the timing is controlled from the software causing such irregularities.

[EL NINO]

Balanced 13 "ultimate works well on the Spectra .. because the Spectra V3's depth and sensitivity depend on the RX gain adjustment level ..- especially in light terrain ..


Equinox has no intrinsic coil diagnostics function ... like Spectra V3, it doesn't even have overload- / except for Beach 1, and Beach2, overloading down its Tx / .... to tell us that the coil may be outside its tolerance ..


According to Equinox, it processes the Rx signal at various gain levels - and therefore does not overload .. even with relatively large metal objects.
The electronics in the coil can do a good part of this work - and if the coil is within the manufacturing tolerance of the balance .. it will work fine according to me ..




Practical tests that I did on 7bar Fe3o4 Black sand -4.4% magnetite - where Equinox worked at almost full sensitivity - sensitivity24 ... tell me that Equinox has a different signal processing RX than usual for VLF detectors ..- Equnox rather, it is governed by a separate RX gain setting (automatically-or implemented in a given recovery speed) and also by a separate sensitivity-user setting


NORDIC .... Of course, I consider your effort to better tune the Equinox coil as a very good step that can improve the detector's performance ... and I'm waiting for your next results ...

[Skippy]

The thing that newbies don't understand about coil nulling is this:
*If the coil is adequately nulled, the detector performs correctly.
*If the coil is super-precisely nulled, the detector performs correctly.
*If the null is beyond some limit, performance is degraded. Typically this would mean the ability to cope with ground that gave a strong signal would degrade, but strong target overload could happen, too. Depending on the detector, sensitivity will drop off, and ultimately fall to near zero.

I have some plans to try out some power line noise filtering, hopefully out in the field, though my break-in cable is a bit fragile .... may need to splash out on a proper free-socket.

[EL NINO]

Mr. Skippy ... all your thoughts are right ..


If the coil balance is within proper manufacturing tolerances, such coil will correctly detect ...


dangerously if the coil is beyond the manufacturing tolerance of the balance.


I agree with the elimination of the Factor Emi I consider the essential element in better detection.

[nordic]

Exactly my findings as well, thanks for clarifying and confirming Skippy. As you have mentioned earlier, the RX signal is biased to about 1.6V via a divider, however a very close large object near the coil will clip the TX signal, I think this is where you see a triangle warning on the equinox as I found out by accident placing the MD on the metal sheet.

Equinox software appears to be looking for changes in the TX signal, so it doesn't really care if there is little TX signal present on the RX as long as it is constant and low enough. Anything not constant, like Vdc noise would have an adverse effect to sensitivity, if I understand it correctly. Same reason why coil needs to be moved over targets and does not work stationary, as it needs changes.

For connector, as you may know already, just look for "M12-8" or "M12 8 pin", but they are quite expensive I have to say...

When I put some filtering on the Vdc, I'm not sure which part it made "better", the power line in the control box or coil circuitry, but brought it down to about 20mV Vpp at the connector. There is background noise and also TX spikes present in it.

[Skippy]

For connector, as you may know already, just look for "M12-8" or "M12 8 pin"
Did you miss my first post?

I just didn't want to spent 15+ Pounds on two connectors until I was sure I would need them. My broken coil already cost too much, no thanks to customs charges and related fees ... the coil cost me 20 US Dollars, I had to cough up 19 Dollars in fees to get it into the UK, plus the expensive international postage charges. Made me mad, partly with myself for not reading all the pages of detailed stuff on the UK government website about charges, plus Royal Mail taking a big cut...sheesh.

[nordic]

Sorry Skippy, my bad, it was last year I had to cough up those as well - curiosity took over.

My plan is power up the coil amp with the battery through a low noise vreg. Separate to the control box. It is probably an overkill, but I want to see what it will result in.