TDI Pro GB Evolution

[simplejohn]

HI All , i've been watching this forom for some time and cannot wait until some of the members here release Gold specific detectors that rival the GPX series.

Until then , i must battle on with the TDI.

Has anyone out there thought up a better way to GB on this machine that could potentially overide existing GB method?
From my limited understanding , the tdi takes a sample for GB then subtracts it from the other samples ? this is what causes it to reduce depth once gb is in use...

i can get my uncle (electronics ENgineer) to work on it but he has not done any work on MD...

GB ala QED would be great but i'd like to avoid going into firmware...

[simplejohn]

Reg are you on this forum? you posted the following on Findall...(note the last line)

Did you go ahead with this magical mystical mod? i would be happy to help (monetary and Testing on AU soil) even if it is in dev stage...(PM Me)


"Reg - July 31, 2011 08:38 AM (GMT)
smdc,

I am not sure why your DD is noisy. I need to do more experimenting with the DD to see if I can figure out what might be happening. I suspect the problem is still tied to your lack of enough GB adjustment.

I have only tried one nugget finder mono and it worked fine. In fact, a friend of mine used to use one on his GS 5, which is basically the same as the TDI in design.

Changing the subject, if you go to the right Whites dealer, I suspect he could make the right adjustment. I would ask to see if he has been trained by the factory and knows just which control does what. It is very easy if you know the right pot to adjust. It is not easy if you don't know which is which and the odds are the wrong one will be moved. If that happens, well.....

Personally, I think the internal GB pot should be An external GB control also. In many areas where the ground is mild, being able to adjust this control allows one the range from almost no GB to full GB. I have this one one of my modified units. So, I can adjust from very little GB to quite a bit more than now. I did limit mine to the worst condition I could find, though.

If I could just get off dead center and back to working on my other ideas, the present GB design could become a thing of the past in the fact there would be no depth loss as it was adjusted. Oh yeah, the side benefit would be the ability to distinguish ferrous objects from non ferrous.

Reg


"

[simplejohn]

Or even an afterburner idea, any thoughts on how?

[Reg]

Hi sj,

Yes I am on this forum also.

The design of the GB on the TDI is the standard designed used by ML as well as that used on the QED. The difference is, instead of one later sample, multiple samples are or could be used. One could also use filtering and less actual GB type sampling.

So, there are other techniques that could be used. The problem lies in interpretation of the patents and what would be upheld if a suit were to occur.

I used to use this filtering idea and less GB in an earlier low powered design and it worked quite well. Since my design was active years ago, I suspect any attempt to patent this concept would be voided.

Eric Foster added additional filtering to his Goldquest and used a DD coil. In many places, this worked well, especially if the operator took care in trying to keep the coil reasonably level.

The problem with the filter and less GB combination is you lose the ability to determine some iron, but on the good side, you gain back some of the depth loss that occurs because of the present GB design.

As to the last line you quoted, I am still working on that design.

Reg

[simplejohn]

Thanks for the response reg. I though there was only a single ground sample and the pot simply amplified it? I might try a range of my own mods based on your posts.
Considering the removal of resistors in the preamp signal, could an afterburner incorporate preamp and tx ? Perhaps a tinkerer style tx might eve do away with damping resistor. i'll need to have a look at a low noise solution to the damping.
also do you know if it is possible to separate gb samples from main signal- this way I can just add a variable delay to these samples. I might just have another look at the circuit. Then the gb can be turned right down.

[Reg]

Hi SJ,

One has to be very careful as to what you do, especially when it comes to the pots and certain resistors or things can get worse very rapidly.

Which version of the TDI do you have? If it is the regular TDI, then is it one of the original 200 units with discrete components or is it one of the newer surface mount models? Neither is easy to work on . The original model is difficult to take apart and there isn't that much room for additional mods while the surface mount is difficult to replace parts or add things.

It is too bad no one has made a simplified kit of the GS 4 because that would be the detector to experiment on.

As for the TDI, a lot of work went into getting it to the stage it is now. Trying to reduce noise is possible but will not display dramatic results and could get quite complex and expensive. Using noiseless resistors is an option but they are expensive plus they are delicate, so one has to be careful with them. Ground balance mods are in the same boat, meaning they are possible but, again, not easy. Making the autotune speed so it is adjustable is even worse, but can be done.

Good luck if you decide to tackle any of the above sections.

Reg

[simplejohn]

Hi Reg,

TDI Pro - Oz , SMD

I am imagining a fair amount of Do and test in small steps. The resistors would be first.

In terms of GB mods, if the ground sample is only taken from the "iron" channel , then could this channel be delayed by X number of pulses to achieve a no loss GB? X being a variable?

I would stay away from GB mods if i could impliment an effective afterburner. The reason is that this would effectively make up for the depth loss of running GB @7-8.
In fact i would do nothing if your design came to fruition... It sounds like it ticks all the boxes.

The other alternative is to sell the tdi and wait for whits new machine...

[Reg]

Hi sj,

Your unit is microprocessor controlled so changing any timing function would require knowing the code and have the ability to properly change it. That isn't a simple task.

Now, with that said, after each pulse a sample is taken for the main sample and one later for the GB sample. Reducing the number of GB samples would only make more GB signal needed and this would cause more noise. More noise, would result in less depth of detection.

I mentioned the GS 4 as a good experimental detector for those wanting to modify the TDI. The reason is, there are a lot of similarities plus each or most timings can be changed. In fact, the SEL 1 and SEL 2 are examples of what will happen if the GB sample is taken at different times with respect to the main sample.

There are two key problems on a PI and one is how and/or when to take the GB sample and the second is how to reduce any noise associated with that sample. The reason for the extra noise from the present GB sample design is the extra gain required for GB to occur.

So, one area to tackle is the noise issue. This can be done on a simple basis or a complicated one. A simple basis is to try to minimize the transfer of noise. Using a separate tx and rx coil design helps with this aspect. So, the use of a DD coil or a concentric coil is a step in the right direction.

Next, on the receive side, you can now use lower noise resistors such as metal foil types. Unfortunately, they are quite expensive. To compliment these resistors, you will need a new lower noise opamp such as the LME49990 or a similar low noise unit.

Of course you could alter the design such that the entire GB channel is designed for ultra low noise including the preamp design itself.

And the list goes on and on and on.

Something as simple as turning back the delay reduces the ground signal significantly in many places. You do lose very small gold, but larger gold should still be easily detected, especially if it is pure. In this situation, you ignore small gold but probably could get more depth on the larger gold. Add a few tricks like using less GB plus filter changes and the depth increase could be quite noticeable.

Finally, one could tackle the noise from a different perspective. This requires studying the present design and altering it such that noise is effectively reduced by simple unique means.

So, there are alternatives. Unfortunately, none are easy to design and implement.

Reg

[Orbit]

I saw how many TDI uses ne5532 input opamp ! whether it is suitable as well as LME49990 which are difficult to can buy in my country .

[Reg]

Hi Orbit,

The standard TDI uses the 5534 opamp. It is a decent opamp and is available in many places. However, there are differences in the 5534 if purchased from different manufacturers. Whether this difference is consistent on a particular manufacturer is unknown. In other words, an amp made in one country may or may not work better than one made in another country.

Eric Foster found this out and posted it somewhere. I am not sure where now. If I remember correctly, the Philip's brand performed the best on his testing.

Now, there are other opamps that work as good substitutes besides the LME49990. One is the OPA211. The OPA1611 is almost identical to the OPA211 but has a greater offset voltage. The OPA228 will function reasonably well as a replacement for the NE5534 also.

Reg

[simplejohn]

Thanks Reg, i'll have a look at some schematics and post if i work anything out....
60 hour work weeks don't help...

[simplejohn]

What if a variable delay is introduced at the end of the iron channel , this would mean that when the gold channel picks up a target it would not be offset by the same target being picked up on the iron channel... this is because of the time delay. this would mean no depth loss . I hope. any thoughts?

[Carl-NC]

John, TDI achieves GB by a proportional subtraction of a ground sample from the main sample. The proportion can be adjusted with delay, pulse width, or gain. TDI fixes the delay and pulse width and varies the gain. If you achieve GB and then vary the delay, GB is lost. In short, this method of GB loses depth and also has a target hole and no manner of adjustments can fix that. It requires a different approach.

[simplejohn]

Hi Carl , great site, just so ive got this right, so the ground sample is taken at a fixed ( different to the dial- longer)delay as target sampling, the gb knob then varies its amplitude then it is subtracted. Going on the gs4 schematic gb is achieved by subtraction of a long delay and a short delay signal, the gb pot simply adds resistance to reduce the short delay signal until gb is achieved. Is it then not possible to increase the delay on the longer delay signal as a means to both reduce the hole and achieve gb ?

[simplejohn]

the below explains a lot...

Long time has gone since I last posted something truly stupid here, so, here it is:
This writing is about one interesting design approach to two sample substraction method, I tried it on prototype with great success. Typical approach, like used in GS detector is two channel, two integrator approach, then substraction is done on integrator outputs. Modified version is with two complete circuits, motion filter included, and substraction done at filter outputs. Highly not recommended however, very high precision and quality\matched components are needed to match response of both channels. I ended up with something else, varying amplifier gain during second pulse to achieve GB adjustment, short description goes like this: Detector is bipolar pulsing (same polarity pulses on bifilar coil, so field is bipolar), no need for EF canceling pulse. Amplifier is differential 2 stage (4 op-amps used) , conveniently AC coupled, differential output too, followed by one single ended integrator. Now, two pulses are used, second one for GB, (exact width, delay and gain change subjected to some experimentation). Timing circuit switches integrator input from one amplifier output during first pulse, to opposite (inverted) one during second pulse, to enable substraction, but amplifier gain is changed after first pulse (to allow settling at new gain), so GB is adjusted by varying (increasing) gain during second pulse. All this in reverse order during reverse polarity pulse. Integrator output will go in one direction for objects below preset TC, in another for longer TC, and remain unchanged for one particular TC (that of ground signal, this is the point of ground balancing). Timing is not complicated at all, only two chips. To disable GB\disc, second pulse is simply disconnected in timing circuit. Not only this achieve nicely adjustable GB, but have some potential with similar design using large bifilar coil and longer, high power pulses. Now GB is not important, but some amount of “discrimination” can be achieved using same method, based on individually adjustable object TC and size criteria. Fortunately, large objects are more predictable in behavior. Aside, bipolar pulsing is very suitable for completely static operation, nice for large coils. This so far works just fine, I hope will have more time soon (my current contract expire end september ) to finish complete design, and publish schematics and all, until then, any opinion or idea on this subject?

You do yourself a injustice as it is a good and practical idea.

I have used a similar scheme for some time i.e. two stage preamp with second stage as a balanced cross coupled differential output with overall gain of 400. two samples in each channel for EF rejection and GB, summing into two matched single time constant integrators, then subtracted in a differential input following stage with gain adjustment in GB channel.

Eric.

I think i may have oversimplified the prior post.

What Eric is doing sounds the business/.... wonder what it would look like on a GS4 or a TDIpro?