Coil shielding.

[Don Bowers]

Hi,
I built Mark Stuarts Microcontroller PI Treasurehunter a while back. One problem it suffers from in my area is a change of tone when the loop is brought close to the ground. I have noticed that this problem is more pronounced when the ground is wet. I played around with shielding using simple aluminum foil
and attaching the end of the shield to the end of one of the wires without much difference.
Any ideas on what might help?

Also, I was wondering if Carls Hammerhead design performs any better than the Microcontroller PI.

Thanks..
Don

[Reg]

Re: Coil shielding.

Hi Don,

Coil shielding is required to minimize the ground capacitance effect, so yes, shielding is necessary if you are going to try to operate at reduced delays.

Now, even with the shielding, there will be a significant ground effect in some areas. The amount will vary with the ground conditions.

So, to answer your question, will the HH design perform any better than the Microcontroller PI? If the question is aimed at the mentioned ground problem, the answer is no. If the question is aimed at overall performance, the answer is yes, the HH is a better overall design and will work better.

The Stuart PI works well and is a good learning tool, but it does have several limitations built in. First, there is no autotune. Without it, there will be drift and retuning will be necessary, quite possibly on a frequent basis. Second, there is no earth effect compensation. As such, any increase in gain or shortening the delay will see an obvious problem with false signals when the coil is swept in the air or across the ground. Third, the power supplies are limited in design. Actually, one is not regulated and this can have a negative impact overall.


Reg

[Don Bowers]

Re: Coil shielding.

Thanks!
Don..

[Mike Kendall]

Re: Coil shielding.

Hi ,
I've built both designs and after tuning the processor ,which I admit took some effort, found the Stuart to actually have better sensitivity, especially to smaller objects. Ground effects
are similar. Best overall performance from my testing, is the Stuart.
Allthough I have a friend who programed the chip for me I'm not an engineer by any means. Just know what works for me.
CYA,Mike

[Reg]

Re: Coil shielding.

Hi Mike,

I have built both designs also and I had to add a form of autotune, plus improve on the power supplies, and I reduced the delay significantly on the Stuart PI to get it to be very sensitive. Once all this was done, I then ran into the earth effect problems. At that point I decided to abandon the project.

Don't get me wrong, the Stuart design is a very good design from which to learn. Plus a lot of things can be done with the program to improve the overall sensitivity.

However, the Hammerhead design improves on some of the weaknesses of the Stuart design with better power supplies, earth field effect compensation, etc.

Now, the sensitivity of the HH can be increased but care has to be taken to build a coil that will allow for the shorter delay to be used. If that is done correctly, then the HH can be built to outperform the Stuart PI quite significantly.

The weak link of the HH has been discussed and that is the audio. It isn't that bad but care has to be taken to make sure the audio doesn't effect the rest of the design.

Actually, if one uses Li Ion batteries at 14.4V and a regulator to regulate the voltage down to 12V or so, then things are greatly stabilized. That is what I ultimately did. This reduces the audio interaction significantly.

Reg

[Carl-NC]

Re: Coil shielding.

Also, I was wondering if Carls Hammerhead design performs any better than the Microcontroller PI.[/i]

Fundamentally, no. The Hammerhead project uses fairly common PI circuitry, same as the Stuart and Proscan designs, and even the same as the Surfmaster PI, Sandshark, and Baracuda. All of these detectors have potentially the same capabilities, if they are set up the same way(coil, frequency, delay, etc.).

Hammerhead, though, has knobs for all of the operating parameters, so it is useful for learning how the parameters affect performance. That's why I designed it the way I did.

- Carl

[Mike Kendall]

Re: Coil shielding.

Hi Reg,
With no engineering background I was merely stating my experiences based on the designs that were available. I obviously missed the discussion you refer to about the HH having a weak link. What I found was high acclaim.
What I can say is that the adjustments designed into the HH with pots and trimmers,are available on the Stuart but uses a reset . Once I tune it to a particular area, it does a decent job,for me, with much more sensitivity. But if you know of any improvements that might help either design, maybe you would share that?
Mike

[Mike Kendall]

Re: Coil shielding.

By the way Reg, I'm assuming that you optimized and rewrote the original hex that came with the Stuart? I had similar results as you described until I had my Son take a look at the program code. There are 4 sub routine errors with the program. I would guess any hardware mods might be futile until the code was corrected.
Here's the errors.

MOVLW 0x00 ; should be MOVWF
TRIS PORTB ; should be TRISB
MOVLW 0x1F ; should be MOVWF
TRIS PORTA ; should be TRISA

If you haven't allready, try that and then check your results.
Mike

[Reg]

Re: Coil shielding.

Hi Mike,

My discussion was not done to criticize either design or anyone but simply to point out that the HH (Hammerhead) is a much better design overall.

I was under the impression from passing emails that Magenta was considering upgrading the Stuart design and add items such as the earth field effect compensation to their new design. I am not sure what happened. Such changes may still be in the works.

Now, the HH design is a good one that provides the owner the opportunity to try several different things very easily to see just how they effect operations. This is what makes this design great.

As for HH upgrades, there has been some discussion previously about this. On thing I mentioned was replacing the 740 FET with a newer design with less output capacitance. This would allow for a shorter delay. One of the newer Fairchild units should work quite well. Somthing like a FQP6N25 will work if the current is limited in the coil to a few amps. This FET is rated at 5.5A, 250V.

The key is the output capacitance of this unit is only 50 to 60 PF where the output of the 740 can be a few hundred PF depending upon the age of the FET. Lowering the output capacitance should allow for a faster decay time and as such a sorter delay can be used.

As for other changes, I have posted several. One other that comes to mind is to change the autotune pot from a 100K to a 1 meg. This gives a much wider range of autotune, which can effectively increase the sensitivity.

A few suggestions have been made as to the settings of various pots also. For example, a 10 usce to 20 usec sample time seems to work best and provides sufficient signal without excessive sampling time which can add noise. The second sample can be done as short as 50 usec later than the main sample, especially if gold objects are the prime objective. Otherwise, maybe a 100 usec wait might be in order. I doubt that one will see much difference between the two.

Ideally, one should try to sample as soon as possible to obtain the best signal from small or gold objects. As such, one might want to keep their coil inductance low, maybe a value of 300 uh or so. This seems to work very well and if all other things are done right, one can achieve a sample time of less than 15 usec or even approach 10 usec. By using something like a 25 to 26 awg wire to add some resistance to reduce the overall current draw of the pulse, and a newer FET, the delay can be shorted much easier.

Reducing the current draw also reduces the negative interactions, otherwise, much heavier batteries should be used, rather than AA's for the best results. Surprisingly, reducing the pulse current and the sample delay time will produce a sensitivity equal to or in most cases, greater than when using a much heavier current and a longer sample time so nothing is lost by making these mods. In fact, I have tried a wide range of pulse times from 20 usec to 150 usec and found that a 50 usec pulse works as well as any for my applications, which is primarily looking for gold objects. In fact, I could see no noticable difference between the 50 usec and the 150 usec settings. So, rather than pulse with a longer pulse, I simply set it at the 50 usec setting and forget it.

As for the audio, this has been discussed before. Because the HH can draw heavy current, and the audio has the potential to do the same, there can be an interaction. In other words, the audio can affect the pulse which then affects the operations. To limit this condition, R 41 should be set for a higher resistance value. Larger batteries can be used also. Another alternative is to add a regulator to the battery supply to minimize the variations. A third alternative is to use an external amp for the audio so R 41 can be set quite high, resistance wise, thus minimizing any interaction.

In summary, both the Stuart and the Hammerhead are good designs and can be quite sensitive. However, the HH is more versatile and allows more "experimenting" to be done. Thus as a learning tool, it has an advantage. The HH design is a better design overall, but one must remember the Stuart design is much older but does introduce the PIC to the PI. So, it has its advantages also.

Reg

[Reg]

Re: Coil shielding.

Hi Mike,

I originally purchased the kit from Magenta and later programmed my own chips. So, I don't know about any errors in their design. Their programming worked but didn't allow for any changes. Third party programs did allow for mods to be done and changes to be made to different settings. However, software changes can't compensate for hardware limitations.

The fact that the Stuart PI doesn't have a regulator on the positive supply does introduce more noise and instability into the design. this is more noticeable if the gain is increased significantly, which I did.

Also, there is no second sample taken to compensate for the earth field effect, so if the sensitivity is increased sufficiently, then sweeping the coil can and will generate a false signal, especially if the coil is suddenly started or stopped.

As for software changes, at the time I was working with the PI, I sort of taught myself via a crash course in programming and generated a very basic program that would allow me to change the delay. That was almost 3 years ago and I haven't done anything with a pic since, so my programming skills are about 0 now. When done, the PI worked extremely well except for the falsing.

Now, I also remember that rocks such as Magnetite would generate a very strong response with the original pulse time of 250 usec and did so until I reduced the pulse on time to less than about 70 usec. I think I finally set the pulse on time for 50 usec to assure I would get minimum response from rocks.

If you don't have any magnetite rocks, you can try a piece of ferrite as a test target to see just how lengthening the pulse can create a signal from this type of rock on the Stuart design.

Detectors with compensation for the earth field effect eliminate this problem of magnetite rocks creating very strong signals. At least, they minimize the problem.

Reg

[Mike Kendall]

Re: Coil shielding.

Interesting reading Reg. Too bad the declarations of most detectors fail to mention the negatives. Be nice to see some actual drawings of variations from different people .
By the way, you failed to mention if you re-assembled the source code when you built the Stuart. Or maybe purchased magentas chip pre-programmed?
I conclude with you on most poiints you make except that I have to stick with what I found to be . That the stuart has far better performance.
Mike

[Reg]

Re: Coil shielding.

Hi Mike,

I mentioned that I purchased the Magenta kit, which includes the pre-programmed chip. So, I started with a programmed chip, but later programmed my own.

I didn't try to disassemble the original source code for the Stuart PI.

The program was extremely basic, considering it only had to pulse for a certain duration and then sample for a certain amount of time.

Initially, I thought I could do the program in basic, but the program was way too slow, so I just intialized in basic and then continued in machine code. This was the fastest and simplest means to write the code for me.

At one point, I started to take a second sample to compensate for the earth field effect, but abandoned the idea when I noticed more problems with instability because of the power supply.

Now, I will admit the simplicity of the design of the Stuart PI, made maximizing the sensitivity quite easy. It is more difficult with the HH but only because one has to become familiar with the various controls to determine the best settings.

As for sensitivity, well, the HH approaches the sensitivity of my GQ clone. It isn't equal but is quite close. I was never able to get the sensitivity of the Stuart close to the other two.

Reg

[ew]

Re: Coil shielding.

Hi Mike,
I saw you mentioned about the PIC CODE correction as below:
MOVLW 0x00 ; should be MOVWF
TRIS PORTB ; should be TRISB
MOVLW 0x1F ; should be MOVWF
TRIS PORTA ; should be TRISA

I download Mark Stuart's EPE projects and code, but can't see the said problem, can you tell me which line? or in which section?
I had just finished that project with PIC16F84 and downloaded code. I havent test run it yet.Wish to correct the code before doing further testing...
Please help.

Regards,
EW