IGSL

[ivconic]

Lo and behold, my first ever video on Youtube:

http://youtu.be/AX6oQ4UDNtk

It is an air test, a short video.
More to come, as soon as I prepare a fixture for a camera on a shaft.

Pretty decent and promising.
It acts as it supposed to act, except response on 5 Kuna coin.
I expected clear high tone response on it but it is responding with mixed tone!?
Maybe you should try to adjust it a bit better so to gain clear high tone on such coin, maybe it is about coil?
The rest; decently and promising!
Cheers!

[simonbaker]

Oh yes there are.
First of all, if your coil is allowed to "float", with unbalanced differential input it will float wildly within the rails voltage. you may check this phenomenon using the model I sent, but to get a more true to the reality picture replace the opamps with some real-world models, like NE5532. The inverting against the noninverting inputs will be fine, but the common mode swing is just fantastic. Center tap fixes that.
There is also a problem of a rig to the coil cable, and the whole shebang being actually an antenna. OK, there is a short piece of cable involved, and problems are not exactly the same as with professional microphones that use matched impedance balanced ... everything. But there is a strong Tx that needs to get balanced out.
In case of a center tap, and aperiodic loading, halves of the coil are not loaded equally, which most probably is not that much of a problem if all the impedances are sky high, and coil is tightly wound. Problems start when you want to achieve low noise AND aperiodic operation AND a center tap. My problem here is that I got ahead of myself and got only half way there. Why aperiodic? Well, I could, within reason, change a Tx frequency when needed, and Rx will not need any adjustment at all.

There is no need for balanced loading in case of high input impedance and periodic loading (a parallel C to make a tank). In resonance a low coil impedance is transformed to a high preamp impedance, while center tap takes care of the common mode. That's what I have right now. OK, common mode balance would fare even better.

VLF radio amateurs gave up on shielding their frame antennas for several reasons, and instead they mostly use center tap and parallel resonance.

...

Your comment from day before yesterday gave me a bit to think. It was one of those "there is no free lunch" revelations. That true balanced opamp configuration lulled me in a dream that I achieved something cool, but in fact only one, and not that important facet was tackled with.
There is a way to deal with CMMR and balanced impedance at once. It is a cross-coupled balanced amplifier. It uses 2 opamps, has symmetrical impedance for both common and differential signal, and huge CMMR. OK, instrumentation amplifier too. OK, and microphone amps such as THAT1510 family if I want to spend too much money.

A cross-coupled preamp could be made as an add-on for the existing rigs, but I'll just leave it at that for now.

BTW, if you observe a typical (R1=R3,R2=R4) configuration, you'll most probably have a capacitor in parallel with R2 to fight RFI ... it also ruins the CMMR. A correct configuration would have the same capacitor in parallel with R4 as well.

To help get it straight in my own mind:

When I say "unbalanced differential input impedance", I'm not talking about removing center tap grounding; I'm just referring to the input resistor values. So I agree the center tap grounded seems desirable to prevent the "float" problem (although dfbowers has achieved some good practical results with no grounding of the RX coil at all).

So, short of a fancy rig like your "cross-coupled balanced amplifier", it seems we can either choose R1=R3,R2=R4 to provide balanced impedance to common mode signals (and lose differential signal impedance balancing), or we can choose the other relationship where the differential signal impedances are balanced, but the common mode signal impedance is not.

I'm trying to figure out if there is any downside to unbalanced differential impedances -- you mentioned aperiodic operation. Would you explain that some more and explain what tradeoffs you see or what the problem is? (I agree with higher resistors hardly would be noticed anyway.)

Reasons for common mode resistor balancing: I'm guessing that "ground charge" (wet grass, etc.) might cause a true common-mode signal that we want to suppress, because it probably couples fairly equally to both sides of the RX coil. It also seems like it could be a high impedance signal, making the input resistor choices more important.

As for the cable being an antenna, usually we use a shielded cable; is it safe to assume that whatever gets past the shield produces a common-mode signal on the RX wires? If so, it supports choosing the "R1=R3,R2=R4" impedance balancing. I guess even if it has differential components, we just can't do anything about it through impedance balancing.

Balancing out the TX signal seem more like a nulling issue rather than a balanced circuit issue to me. Perhaps there is a common mode (capacitive) link worth addressing?

I played with some sims and noticed that if I used "common-mode" impedance balancing (R1=R3,R2=R4) and low impedance input (R1=R3=68ohm, R2=R4=3k3), the RX signal dropped extremely low, probably because the inverting input has very low resistance (approx 34 ohm) to the diff signal. So I conclude that if resistors are chosen for "common-mode signal impedance balancing), you can't make them to low-valued. This is probably obvious.

Thanks for you video. If you can make a schematic of your MD, it would be very helpful to look at.

Regards,

-SB

[Davor]

Lo and behold, my first ever video on Youtube:

http://youtu.be/AX6oQ4UDNtk

It is an air test, a short video.
More to come, as soon as I prepare a fixture for a camera on a shaft.

[Davor]

Oh yes there are.
First of all, if your coil is allowed to "float", with unbalanced differential input it will float wildly within the rails voltage. you may check this phenomenon using the model I sent, but to get a more true to the reality picture replace the opamps with some real-world models, like NE5532. The inverting against the noninverting inputs will be fine, but the common mode swing is just fantastic. Center tap fixes that.
There is also a problem of a rig to the coil cable, and the whole shebang being actually an antenna. OK, there is a short piece of cable involved, and problems are not exactly the same as with professional microphones that use matched impedance balanced ... everything. But there is a strong Tx that needs to get balanced out.
In case of a center tap, and aperiodic loading, halves of the coil are not loaded equally, which most probably is not that much of a problem if all the impedances are sky high, and coil is tightly wound. Problems start when you want to achieve low noise AND aperiodic operation AND a center tap. My problem here is that I got ahead of myself and got only half way there. Why aperiodic? Well, I could, within reason, change a Tx frequency when needed, and Rx will not need any adjustment at all.

There is no need for balanced loading in case of high input impedance and periodic loading (a parallel C to make a tank). In resonance a low coil impedance is transformed to a high preamp impedance, while center tap takes care of the common mode. That's what I have right now. OK, common mode balance would fare even better.

VLF radio amateurs gave up on shielding their frame antennas for several reasons, and instead they mostly use center tap and parallel resonance.

...

Your comment from day before yesterday gave me a bit to think. It was one of those "there is no free lunch" revelations. That true balanced opamp configuration lulled me in a dream that I achieved something cool, but in fact only one, and not that important facet was tackled with.
There is a way to deal with CMMR and balanced impedance at once. It is a cross-coupled balanced amplifier. It uses 2 opamps, has symmetrical impedance for both common and differential signal, and huge CMMR. OK, instrumentation amplifier too. OK, and microphone amps such as THAT1510 family if I want to spend too much money.

A cross-coupled preamp could be made as an add-on for the existing rigs, but I'll just leave it at that for now.

BTW, if you observe a typical (R1=R3,R2=R4) configuration, you'll most probably have a capacitor in parallel with R2 to fight RFI ... it also ruins the CMMR. A correct configuration would have the same capacitor in parallel with R4 as well.

[simonbaker]

You are right, of course. There are always some tradeoffs. That is the reason I wound the Rx coil bifilarly with a center tap. This way my common mode response is set by a coil resistance, rather than the component tolerances.

My intention was to reduce noise as far as possible and give the coils a spa treatment at balanced loading, but now when you mention that, and my recent experience with a real detector - I might reconsider this true differential idea. External noise by far exceeds the thermal noise, so finetuning noise makes little sense. Instrumentation amplifier is not optimal choice when noise reduction is of essence, but heck, it works well.

Anyway, I am attaching two LTspice models, which are basically one model but with worst case scenario in one, and Monte Carlo scenario in the other. Both are done with 1% resistor tolerances and both of them indicate loud and clear why in classic design you really need coil shielding.

enjoy

Thanks for Spice models, interesting comparisons.

However, I never had problem with the center tap -- it was just the resistor values and whether you choose to balance the differential input signal or the common mode input signal that was the question.

In your worst case example, if you put back the center tap on the "common-mode balanced" circuit (R1=R3,R2=R4), then that case survives the resistor variations best of all, it seems, so still would be my choice there. (I'm still trying to get a better understanding of the various benefits and effects of the grounded center-tap).

But your idea of lowering the input impedance of the amp stage seems like a reasonable way to reduce noise (although it seems a noise analysis would be useful to determine where the optimal point would be, because lowering the input impedance will drop the coil signal, and if the op amp contributes significant noise, you may lose S/N if you go too far).

So given your goals, why not choose a common-mode-balanced set of resistors that are also low impedance (68/3k3, 68/3k3)? Is there any reason to want to balance the input impedances for the "differential" input signal?

-SB

[Davor]

You are right, of course. There are always some tradeoffs. That is the reason I wound the Rx coil bifilarly with a center tap. This way my common mode response is set by a coil resistance, rather than the component tolerances.

My intention was to reduce noise as far as possible and give the coils a spa treatment at balanced loading, but now when you mention that, and my recent experience with a real detector - I might reconsider this true differential idea. External noise by far exceeds the thermal noise, so finetuning noise makes little sense. Instrumentation amplifier is not optimal choice when noise reduction is of essence, but heck, it works well.

Anyway, I am attaching two LTspice models, which are basically one model but with worst case scenario in one, and Monte Carlo scenario in the other. Both are done with 1% resistor tolerances and both of them indicate loud and clear why in classic design you really need coil shielding.

enjoy

[simonbaker]

ref

...check the attached PDF document and seek "The True Diff Amp"
Works nice in IGSL with values as per the quoted post.
It is my solution to shielding a coil ... without shielding a coil.

Davor:

It seems to me your choice of resistors provides equal input inpedance for the differential voltage, i.e. any signals picked up by the center-tap-grounded RX coil (including target signals and "null" signal), and is "balanced" from that standpoint.

However, the input impedances are not balanced for common mode signals. It seems to me the whole point of a balanced, differential amp design is to cancel out the common mode junk.

I believe the original resistors provide equal input impedance for common mode signals, and therefore would be the correct choice for reducing common-mode noise.

I don't know if in real-world practice it makes much difference though. My feeling is the RX coil noise (differential) dominates any common mode cable noise, and that can't be removed by any balancing. However, wet grass static could perhaps be a large, common-mode, high-impedance signal and the original resistors could be important there.

Anyway, if we stay with the original resistors, a slight adjustment to the center tap location might be resorted to for balancing the differential signal, but probably not an issue.

-SB

[ahmad]

where can i find last pcb & ... documents?

[Davor]

I must reconsider my thoughts on a noise blanker feasibility in a post #1359:

Electronics tinkering way around this would be a noise blanker, but it would require either a tuned tank with high Q (impractical) or a delay line to be effective.

I realised that all the ingredients for noise blanking are already there, except the noise gate, and it may require a bit of tinkering. I have left an unused opamp on the board so it might become a noise gate - not sure how. Anyway, we have a bandpass filtered Rx input - that will be supplied to the noise gate input. In the gain stages we also have at least two stages of LPF and exactly 2 stages of HPF that form a much narrower band pass with uniform phase change, hence a steady group delay. So I don't need an extra high Q filter - I have it already, and it provides a nice delay. I just have to apply noise gate pulses to the Disc. sens. control, and voila - a noise blanker.

OK, that may require a tweak or two, but I think it is feasible and may be realised incredibly simple. And puff goes the chatter.

...

We have Bora blowing hard again, and my shack under the stars is closed for business while it lasts. When it stops I'll finally make the promised videos (I hope).

[simonbaker]

...check the attached PDF document and seek "The True Diff Amp"
Works nice in IGSL with values as per the quoted post.
It is my solution to shielding a coil ... without shielding a coil.

Thanks.

-SB

[Davor]

ref

...check the attached PDF document and seek "The True Diff Amp"
Works nice in IGSL with values as per the quoted post.
It is my solution to shielding a coil ... without shielding a coil.

[simonbaker]

It is funny but making a true differential input requires just adjusting a few resistors values - nothing else. It goes like this... The inverting side resistors R1 and R2 set the gain, and the noninverting side resistors just divide the input voltage so that R1/R2=R3/R4. It is a common practice to use R1=R3 and R2=R4 because in ancient times of lousy opamps it was the only way to keep the output offset at bay. Even today with bipolar opamps you get minimum offset this way.

In case of AC coupling to the next stage (✓) you can freely play with values R3 and R4, and the worthy goal is to achieve a symmetric impedance to both inputs against the ground.

By virtue of feedback the inverting input impedance of symmetrically fed opamp is ~R1/2, and at noninverting side it is R3+R4. Matching these gives you a true differential preamp. With a little offset, of course.

I am using 6k8 330k 68ohm 3k3 values to get amplification of 48.5 and impedance mismatch of ~2% (which is within the component tolerances anyway). My total input impedance is ~6k8, while the original IGSL is over 220k. I expect some noise improvement this way, but also somewhat more demanding GEB tuning range - for which I know the cure.

ref

[simonbaker]

Hi Jerry, they are one and the same.

The Rx is a zero or low IF type, our target signals after the mixer/sychronous detector - reside at anywhere from 0 to 15Hz and is dependant on shaft length and the demeanor of the operator!

Ive done some club digs in the UK with 80 people in a line up with a shotgun start - the front runners will cut you in half with their mad hacking slash and fully extended units.

It would be real neat to get a sig gen on these units to get a real number for performance - rather than where we are now.

It may be possible to take a sniff of the Tx coil signal, buffer it to provide isolation, add some FM somehow, attenuate to uV range and wire to Rx input.

If your Tx was derived from a xtal you could use a quality synthesised signal 9Hz above or below to give an IF signal in the Filter bandwidth.

Or maybe a VCO phase locked to your Tx with Audio tone as a modulation input to FM it and feed that to RX.


If the Rx sig was simply phase shifted from the Tx (which it isnt its frequency shifted also i.e. more than a cycle of shift) you could have put the Tx signal down a roll of coax to get a phase shifted signal for the Rx, or another RC delay.

S

Interesting parallels . How do you conclude there is more than a cycle of shift, at the frequency we typically use (14.5 kHz)? I don't see that. However, I have wondered if we used a much higher frequency, would we see greater shifts? Simple question to answer with LTSpice maybe -- it depends though on the target model you assume.

But using a different IF frequency to beat against (if I understand what you mean): it seems it would be even harder to detect the phase shifts due to targets. It would be good for frequency modulation of the TX signal, but targets are not supposed to do that (although they probably do -- but probably hard to discriminate effect of small ferrite vs large conductive, etc.).

Good stuff to think about though.

-SB

[Qiaozhi]

Ok but why give this value of 16.1khz on RX as a "reference"?

and according to the formula F0 = 159 / (R (6500X15000)) = 0.01610Mhz = 16.1Khz

Because this is the frequency to which the RX coil is tuned. Note that this is not the frequency that is being received. You will still receive the TX frequency, but at a lower amplitude.

The RX coil is tuned off frequency for two main reasons:
1. To achieve the correct initial phase-shift required by the synchronous demodulators.
2. To increase the stability of the detector and to make it easier to manufacture in a repeatable fashion.

[Davor]

I think a 4 quadrant Rx with a differential Rx coil will fix it permanently. I think the "8" encircled with a Tx will perform just as a DD, yet with EMI suppression, and a great deal of ground suppression too. Besides, such Tx coil will have somewhat better Q and might allow for deeper detection as well.
I'm collecting concepts to pull it through, and it just might be an IGSL add-on at the end. The more I think about 4 quadrant, the more I miss it.

BTW, I went a bit hunting today, and found a piece of thin walled cast iron bowl, or a helmet or something. Can't know for sure till I find the rest of it, can I? Digging was murder. Dry and hard as concrete.
I'm very happy with double tone for aluminum foil.
By curiosity I also visited a site that archeologists gave up on recently, and found many pieces of, well, iron left just underneath their hole. I did not dig there. Guess they were paid to dig 40cm, so when they did - they left.