Few general questin-opinions

[suhoj]

First of all-I apologize for my bad English!
There are several issues that I would point out!
-Good coil MUST have good Q-factor!But-I was unable to find single word about Q-factor for RX/TX coils...
What about old rule:acceptable Q,we shall get if C has capacity in pF as wavelength meters!For example, 1000pF for wavelength of 1000 meters.It is not best possible Q but if we do not have abilities for Q measuring,that is good enough!
-If we have problem with coil nulling (TX signal canceling in RX coil),WHY TX unit is not screened or shielded?For good separation between TX/RX,TX unit should not be even on the RX PCB,in effort to minimize influence of TX signal on RX circuitry,would be reasonable to put TX circuit in separated PCB,and then make good shield in form of metal box!By doing so-one can be sure,that there is no TX signal coming from unvanted coupling between PCB tracks or component mounted on PCB!
-I have no illusion about noise!Since we are not talking about space communications,we are free to forget about noise in kHz spectrum!
So far-this is enough!Any opinion-suggestions about mentioned issues are very welcome from me.
Thank you in advance!

[mikebg]

First of all-I apologize for my bad English!

Suhoj, yor English is not so bad for me, but you can use mashine translator and spell check software. However your question is for other section. You know enough English to see that there is other section for coils.

-Good coil MUST have good Q-factor!

Why? At wideband metal detectors we use damping resistors to decrease the Q-factor. High Q-factor of coils increases problems at narrow band metal detectors.

But-I was unable to find single word about Q-factor for RX/TX coils...

There are different single words in TIME DOMAIN and in FREQUENCY DOMAIN.
In time domain, the word is "Design parameter" or "Coil timeconstant" L/r. The ratio has dimension seconds. A designer should use minimal allowed design parameter because a coil with large timeconstant is expensive, weighty and volumenous because need more metal. For minimal weight is used aluminium.

The word "Q-factor" belongs to frequency domain. This is coil timeconstant multiplyed to angular frequency. The parameter 2пL/r is dimensionless (like relative operating frequency). Note that when we increase operating frequency, the Q-factor increases.

Avoid to use high Q RX and TX coils even in narrow band metal detectors because the ground operates as a core. For RX tank circuit this makes at resonance phase instability (the discrimination is not stabile). For stabile discrmination, designers use resonance frequency of RX tank circuit different from TX frequency. Don't think for noise in RX input. The preamp circuit generates more noise than the resistance of RX coil.

For TX tank circuit, a low coil resistance increases energy efficiency. However the efficiency is decreased because the ground absobs energy, The ground imports a virtual resisrance which is added to coil resistance. When the coil resistance is relatively small, the amplitude modulation caused by ground is increased. You should use a P-I controller to suppress the modulation of TX.

[Carl-NC]

Mike's response was good, I'll add a little more. Detectors use air-core coils so we're not very concerned with Q. In narrowband detectors the TX coil is usually resonated with a C but even then a high Q would make things squirrelly so we avoid that. The RX is kept low-Q to make phase disc reliable.

On nulling, the mechanical coil null is typically -60dB, and coupling from the TX circuit into the RX circuit can pretty easily be kept below this. Just be careful with the layout and power domains.

[suhoj]

I am sorry to say,but I shall disagree with some of statements issued here!
First of all-thank you for prompt answers/reply!
About Q in term of frequency-as frequncy goes high,good Q is far from high,but,it is not the point!Let me say that I am in electronic business,in RF,mostly in microwave region.I started interesting for MD last few years,just for fan,and I shall build one for my self.Hence,I started to "Googling" in order to see what is going on.I prefer discriminate mode so I decided to study available designs and approach!And,as far I have seen-they are all most the same,at least in approach that I am talking about!In order to not argue with any one,I shall try following this criteria:
Since we have to receive signals from the buried objects,in my opinion,we need good Q in RX coil,and reason is very simple.High Q means high received amplitude and bigger deepth accordingly.Let me explain my "philosophy"!We can "translate" phase shifting in frequency change,of course,in very specific manner!So,we must avoid any influence on TX frequency!In every design I have seen,TX coil is part of oscillator it self,because of that-it is impossible to avoid frequency change if we have in vicinity of coil metal object of any kind!In order to avoid that,I shall try to make TX from few stages,first shall be oscillator,and final stage shall be power amp feeding TX coil,so,object of any kind can not frequency change.
Few words about RX:I shall try high Q because of higher induced voltage.RX coil is very first device in receiver chain,so,regardless of my good will,I can not see reason,why signal level in RX coil ,should be deliberately ignored by lowering Q of RX coil!Further amplification in RX chain does not make many sense,if we missed maximum level form RX coil it self!Further more,if we dealing with levels of let say -60dB or so,it is disaster to construct TX and RX on the same PCB!TX must be in my opinion separate unit,well and very carefully shielded,so that TX signal can only appear in TX coil!
Signals processing,interpreting and coil nulling are different story.But,I am sure,"battle" for ability for deeper search-must be won in very first stage of receiver,and,with out TX leakage on the PCB!
I can agree that this could be a little different approach in MD design,but,in my brain,we can not avoid rules and laws that are long time ago valid for any reception.
And finally and again,I expressed my opinion,with out intention to argue with any one,so,any constructive discussion is very welcome!

[dfbowers]

Suhoj,
I think Mike and Carl's explanations are outstanding but I can site an example where a manufacturer intentionally offers a low "Q" coil to deal with some ground conditions.
Keep in mind that the soil type is most often the limiting factor and your design efforts may soon be focused there. You may be horrified to find out that the performance of your new design in practice is only 1/2 of what you get in theory. Not trying to discourage you, but that's real world.
Don

[WM6]

,it is disaster to construct TX and RX on the same PCB!TX must be in my opinion separate unit,well and very carefully shielded,so that TX signal can only appear in TX coil!

Probably you forgot that TX and RX coils are inductively coupled, so shielding of TX stage can be of little use.

[Carl-NC]

Suhoj, the best I can say is, "Try it and see." And since that seems to be what you are planning to do, we'll wait to see what you find out.

[Skippy]

Suhoj. As part of your research, you should examine Nexus detectors, which use a tuned receive circuit. They appear to work OK, but almost all other machines use much lower Q receive circuits, with great success. Also, many modern detectors (Teknetics T2 as just one example) and some older types (Fisher 1266 etc) use crystal oscillators & dividers to define the transmit frequency, and then use a power amplifier stage to drive the tuned transmit coil.

[Qiaozhi]

Suhoj - I understand your thought process. Many others have followed this same route, and you are absolutely right in planning to do your own experiments. That's the only way to get a proper understanding of the problem, as (with metal detector technology) the real world has a nasty habit of screwing up many people's pet theories. Skippy mentioned Nexus detectors (www.nexusdetectors.com) who are the only MD company to use resonant high-Q coils in their designs. There are pros and cons to this approach, and no doubt you'll discover these in your research.

Good luck, and please keep us all informed of your progress.

[Davor]

My 5 cents ... High Q is OK in Tx if you can handle it.
Rx requirements are mostly set by noise and intermodulation limitations. Low noise, counter-intuitively is not achieved by high Q because a parallel Rx tank is in fact a 2nd order low pass filter, so the bandpass high Q reasoning does not apply.
Using a series resonance is OK because it is a bandpass and it reduces the offending 50/60Hz mains hum, but low noise design requires that the shunt impedance of a preamp input is significantly higher than the source resistance, so the resulting Q is on a low side.

[mikebg]

My 5 cents ... High Q is OK in Tx if you can handle it.
.......

Davor, Does this mean that you now recognize the need for P-I-D controller in TX?

[Davor]

No. It just means that I know how to handle it. PID controllers in oscillators lack stability, and regulation by changing conduction angle is wrong. The worst part of PID is the "I" as it introduces relaxation instability to Tx amplitude.

[mikebg]

No. It just means that I know how to handle it. PID controllers in oscillators lack stability, and regulation by changing conduction angle is wrong. The worst part of PID is the "I" as it introduces relaxation instability to Tx amplitude.

This is not true. Make a measurement to see what happens at high Q-factor of TX tank circuit. Here is the principle of measurement:

http://www.geotech1.com/forums/showt...130#post168130

and here is a method to avoid movement and use of salt water:

http://www.geotech1.com/forums/showt...476#post105476

[suhoj]

Thank you for replys gentlemen...thanks a lot,too many issues to reply...
I shall not put in doubts integrity of any forum member.But in same time,I shall say,that it is wise to consider technical statement that exists long time ago,and as such,became a bible of all electronic technician and engineers,I think all of as shall agree with that!
-About high Q in TX coil:TX coil must be well matched to TX output regardless of its type!If there is no match,then,we shall have mismatched feeding which in turn cause standing waves,and those waves in turn are going back over TX cable shield,and as result,we have unwanted TX signal on the PCB of MD,with unpredictable behavior in terms of radiating and even worse,such signal can easy reach RX front and,so that RX can suffer loss of sensitivity,and wrong results reading/interpreting in further circuitry.In my opinion,it is mandatory that TX coil should be well matched to TX output!It is essential condition in TX use.
-In regard with crystal oscillator and divider,with to day components,it is pace of cake to create LC oscillator with high stability and with low phase noise(no one has mentioned phase noise in TX osillator?!)
-About shielding,only I can say,try to see TX signal on the output of RX front and with out RX coil connected.I think every one shall be surprised.We talking about levels of minus 60dB!!!It is not even easy to measure,but,for just few dB gain or difference,people are doing tremendous efforts in term of circuit complexity and/or components used!3dB is often "just" half of something,I am the man who shall not refuse to take advantage of so "small" piece...
-Noise again:I believe and I am convinced that to day components are too good for kHz region!If someone is scared with noise,I have advice-do not connect any resistor parallel to RX coil,since because of resistors behavior-noise shall become higher!I think that it is better to think and consider steps in order to improve signal to noise ratio in RX>But,this ratio shall not be improved by lowering RX signal in RX coil,not at all...Signal to nooise ratio is crucial aspect in to days receivers quality!MD needs good RX!
-About factory made MD circuits I shall not waist my and yours time!I would only say,that I believe they far ahead in research than they put on the market!We have to understand that because of a lot differential commercial reasons,they are not in hurry,as long as they selling products...
-There is also one thing I have not mentioned earlier,but,I believe it is worth to discus.Nulling!It can be done by placement of RX/TX coils in terms of theirs interactions!But,can we go little further?I think we could!What if we introduce negative feedback in RX front-and,in attempt to kill TX signal totally?In to days electronic devices we have such circuitry in regular use few decades ago,so why not?
So far,I have whole concept in my head,and in moths that are coming I shall try described ideas.All my professional works I doing at my home,so that instruments are in hand range all day long!I shall make only TX and RX front-and,and basically I shall be focused on sensitivity only.Readings interpreting and evaluating are different story,and there are a lot very intelligent idea and solutions that are easy to implement.
Thank you for sheering your ideas,remarks and opinions,I very appreciate that.

[WM6]

So far,I have whole concept in my head,and in moths that are coming I shall try described ideas.All my professional works I doing at my home,so that instruments are in hand range all day long!I shall make only TX and RX front-and,and basically I shall be focused on sensitivity only.Readings interpreting and evaluating are different story,and there are a lot very intelligent idea and solutions that are easy to implement.
Thank you for sheering your ideas,remarks and opinions,I very appreciate that.

Hi suhoj,

thank you for sharing your interesting ideas with us. I hope you will be back soon with some working prototype.