All sorts of noise at the preamp.

[Ferric Toes]

Hi Carl,

In the equivalent schematic shown on the 5534 data sheet, the pair of back to back diodes are directly across pins 2 and 3. If pin 3 is grounded, then the clamping should be 0.6V either way. Same as external diodes would. However, there is a peak current limit somewhere, that is less than for external 1N4148's.

Eric.

[Carl-NC]

OK, didn't realize (or remember) that... most opamps probably don't have diodes across the inputs. Hmmm... I now vaguely recall that this has been brought up before... memory's not what it used to be, which wasn't much even then.

[Ray-NM]

Here is one of my what if's.

I have not posted in a long time just read and try to copy what all of you suggest. Any way here is my Idea. Probably been thought of and rejected by others in the past. To decrease cable capacitance you should use one with the lowest Capacitance as possible or a shorter cable . Why not move the electronics down the shaft half way then run the controls you need up a cable nearer to the handle you have just decrease the cable cap by half and decrease the time to sample. I know others have suggested placing the electronics in the coil housing but then you have a long battery cable to contend with. Just a thought. Go ahead and correct me I can take it.
Ray-NM

[bbsailor]

What If? A True Trade-off Analysis.

Ray-NM

You pose some good points to consider. Lets say we take an inventory of all the things that contribute to the coil circuit capacitance that drives a particular value of damping resistor. Then we can see which changes are the easiest to make and also tend to introduce less other problems that must be compensated.

1. Coil capacitance based on the insulation type, thickness and total number of coil turns.
2. Capacitance of the coil shield based on the thickness and dielectric constant of the shield-to-coil spacer.
3. Length and type of coax wire used to connect the coil to the control box. With using coax with about 25 to 30 pf/ft, you will have a total coax capacitance of 65 to 75 pf for a 30 inch coax. A 30 inch long coax will work nicely with a 6" long control box and you may be able to shave a few pfs more off the total cable capacitance if you are of shorter stature and can get away with a slightly shorter cable.
4. MOSFET capacitance can range from 700 to 1000 pfs for older MOSFETS to 35 to 70 pfs for more modern low COSS versions.
5. Circuit capacitance and loading. Here the input resistor to the first stage op amp that sets the gain with the feedback resistor actually appears in parallel with the damping resistor across the coil only while the flyback spike is approaching zero. This previous point is a tidbit from Eric Foster. Therefore the effect of changing the damping resistor has less effect when you consider that the typically 1K ohm input resistor is in parallel with the damping resistor. So in this case you would need to reduce the coil/shield/cable/MOSFET capacitance substantially to see any effect more than a hundred or two hundered ohms. Increase the input resistor to 2.2K to reduce gain (and improve bandwidth and recovery time), and any changes in coil/shield/cable/MOSFET capacitance will be seen as a larger variance in the damping resistor value.

Now for the trade-off analysis. Given that the best decrease that could be achieved with the coax alone would be 60 to 75 pf, what other trade-offs could be done easier to achieve the same or better results? I would try a new low COSS MOSFET, a Teflon insulated wire coil (300 uH), AWG30, 5 ohms, a Polyethelene (PE) spiral wrap vice a PVC spiral wrap due to the lower dielectric of PE, a shield that is not detected by the coil or is minimally detected by the coil a low delays.

The assumption is that the reason why you seek lower capacitance is to improve sensitivity to low conductivity targets. Then when seeking to eek out the last uS of sampling speed, I would lower the control box closer to the coil, but not so close that I introduce other problems. Then I could knock off the last 30 to 37 pfs in an attempt to sample a fraction of a uS sooner and raise the dampling resistor to it's highest critical damping value, indicating that I have less capacitance stored energy to damp.

I would be interested to hear other opinions on approaching this trade off analysis.

bbsailor

[Ferric Toes]

Hi bbsailor,

If cost is no object, you can buy PTFE spiral wrap. It would be interesting to see what difference it makes. I have hesitated so far because of the cost. It would be nice to get a sample from a supplier before committing.

It is surprising how costs would mount up on a simple mono coil with PTFE insulated wire, PTFE spirowrap, copper braid shielding tape etc. The latter item costs me nearly £40 ($73) for a 50m roll.

Eric.

[Tinkerer]

Now what is wrong?

I have not posted in a long time just read and try to copy what all of you suggest. Any way here is my Idea. Probably been thought of and rejected by others in the past. To decrease cable capacitance you should use one with the lowest Capacitance as possible or a shorter cable . Why not move the electronics down the shaft half way then run the controls you need up a cable nearer to the handle you have just decrease the cable cap by half and decrease the time to sample. I know others have suggested placing the electronics in the coil housing but then you have a long battery cable to contend with. Just a thought. Go ahead and correct me I can take it.
Ray-NM

Thanks for all the excellent advice. Lots of things to try and lots of fun trying. So I reduced power, from 12 Ohm total resistance to 32 Ohm. reduced the delay by halve a uS. to 4.5 uS Then changed the cable for RG 58U coax, totally bad. changed for other piece of coax , still bad. Changed to two legs of ordinary magnet wire, back to the same good result as with the shielded tween cable with a lot of noise added as expected. So what is wrong with the coax setup now?

Tinkerer

[bbsailor]

Eric,

I found that the full coil-to-shield capacitance is not imposed on lowering the coil's self resonance. Only about 25% of the coil-to-shield capacitance is actually imposed on the total coil capacitance thus lowering the coil's self resonance slightly. Therefore, based on this characteristic, paying for the absolute lowest dielectric coil spacer is probably not worth the PTFE investment and the benefits might not be seen in the final product. It would be good to make one coil, pulling out all the stops and optimizing each potential variable, just to have a reference coil, fully documented while it is being built, and see where the most practical improvements could be made.

PTFE insulated coil wire is the best and most practical improvement that PI coil builders could use to improve a coil, followed by a low COSS MOSFET with a low peak coil current and higher pulse rate. Ebay has some AWG30 PTFE (Teflon) wire for about $20.00. 500 ft of PTFE wire can make about 10 PI coils 10.6" OD, 18 turns, 300 uH.

Eric, have you ever measured the amount of capacitance reduction necessary to sample 1 uS faster?

bbsailor

[Unregistered]

Hi Tinkerer,

Haven't we chatted somewhere before?

It seems like I recognize some of the info you have posted here.

Ok, one thing I have found out is if you can sample at about 7 usec to 8 usec, you can detect about as small of gold as you might want to find. So, you are at the point now that will work quite well.

Quite some time ago, I found a 1.5 grain (thats grain, not gram) nugget with a delay of around 9 usec. Granted the signal wasn't an ear blaster but was loud enough that I could hear it at a depth of about 1" or so.

Now, drop the delay down a couple of more usec's an things do get better. Even the so called "invisible" nuggets found by John Blennert southeast of Tucson become visible. Personally, I have found the delay of about 7 to 8 usec to be more than short enough to detect most of those types of nuggets.

The point I am trying to make is looking at your waveforms, you should be able to operate at the range you are at now and find very small gold.

Just as important as reducing the delay is the reduction of as much noise as possible. So, try to use the best techniques possible in this area also.

It is great to get the delay down to 5 usec or so, but if the rest of the circuitry isn't fine tuned, then the short delay won't be nearly as effective.

Don't be afraid of building a low powered unit. They really are much more effective than people realize.

Now, as for coil cable, I prefer the SVHS cable as mentioned by Eric. The shielding is such that is isn't detectable like it is on some of the different coax used. I like the Mogami type but do wish it had a second layer of insulation like most of the S video cable. At about 17 pf per foot, it doesn't add much capacitance.

I think you will find that the FET is the dominating source of the capacitance problem. Even changing to one of the newer lower output capacitance type FET's will still produce quite a bit of capacitance. Figure out a way to reduce that capacitance and you will have a winner.

BBsailor and Eric are right about the Teflon insulated wire as the best. This is especially true when building coils near the 11" diameter and larger range.

Now, as you have found out, reducing the capacitance may not be the ideal setup after all. Once you reach that "sweet spot" as you called it, the interaction between the transmit and the receive are critical. Change anything and the "balance" is upset and you will have a difficult time getting back to the ideal signal. In other words, reduce the capacitance and you just might see the decay take longer to settle down.

This is frustrating but solvable. Just remember that you need to sample when the signal is stable. Anything before that is of little value, in my opinion. Granted, it is nice to have the perfect decay curve, but that becomes almost impossible, especially when trying to build a unit that can use many different coils and operate well below 10 usec. So, don't be afraid of using a decay curve that doesn't look perfect. Once again, as long as things are stable at the time you sample, then all should be ok.

Reg

[Unregistered]

JC1

Hi Tinker,

I didn't read all these posts so this may have already been said, put a small capacitor 1-5 pf around your feedback resistor on the 5534 to help cancel the input cap of the op amp, may help prevent ringing.

[Unregistered]

JC1

http://focus.ti.com/lit/an/sloa020a/sloa020a.pdf

[Tinkerer]

Many thanks to all

Thanks to all.


Thanks to all the help received, I got the delay down to 3.5 uS. I will stop here and finish the coil, pot it and shield it and then see where it ends up.

Here are the parameters: Diameter 11’, resistance 2 ohms of the coil +30 ohms, 310 uH, 23 turns of 0.6 mm diameter magnet wire + 7 ft of two conductor shielded cable.
Tx at 65 uS, 12 V, Mosfet NEC 2141, Damping resistance of 626 Ohm, preamp NE5534, with 10k input resistor.

It is impressive to see how much increase in signal strength is obtained by the short delay.

From the above you can see that there are lots of possibilities for improvement left. One thing I would like to try is a discrete FET amplifier before the NE5534 or a low noise, high speed FET input preamp. Could somebody help me with suggestions for that?
I have bought a SVHD cable to try, but have not done the test yet.
Looking for better choice of Mosfet also, the above setup is drawing little power.

Much improvement needed on the power supply, is very noisy, with much feed through from the digital timing.

The timing at present is just a makeshift setup on the bread board, hope to eventually change that for a PIC877 Microchip. Is there somebody willing to help me with that?

Again, many thanks to all, specially Reg, Eric, Carl, bbsailor, ICL, pml and all the others who have supplied valuable information.

Tinkerer

[bbsailor]

Tinkerer,

Most of what I learned came from Eric and Reg. I read each post on Eric's Findmall forum, as well as followed many of Reg's tips based on his quest for seeking the optimum gold PI machine. Carl's Geotech forum is also an invaluable source of technical information and opportunity to share what we have learned. I also wound many coils of different insulation types and shielding methods, noted the differences and applied this to optimizing my CS6PI, Eric Foster-designed machine.

Please post your schematic for the 3.5 uS circuit.

Once you shield and pot your coil some things will change. Most likely, your delay will be extended slightly. Let us know?

Why use the thick 0.6mm wire along with a 30 ohm resistor when you can use AWG 30 PTFE (Teflon) insulated wire and have a lighter coil that is 5 ohms and use a 27 ohm series resistor? Try using 18 turns of AWG 30 Teflon wire and I suspect the damping resistor will be about a hundred ohms (or more) higher indicating a lower coil capacitance.

For the FET preamp, look how microphone preamps are phantom powered. This way, one pair of wires coils allow you to embed an FET preamp (gain 10 to 20) in a small module (point to point solder the parts that make up the FET preamp circuit and encase it in epoxy) to be located in the coil housing along with the clamping diodes. Minelab built a preamp into their DD Sovereign coil housings to improve the S/N ratio. Double D coils with a built in RX preamp are, in my estimation, the next wave for improving the sensitivity in PI detectors. Just make sure you use circuit parts that are miniminly detected (don't use aluminum can electro. caps).

bbsailor

[Tinkerer]

Shielding the coil

Hi bbsailor,

I agree with you, many things could be improved yet. Here is my shielding progress:

Picture shield #1, coil 310 uH, 32Ohm total resistance, Tx 15uS, 12V, scope set at 1uS division
#2, coil placed in shielded housing, shield not connected
#3, connect shield to shield of cable
#4, adjust damping resistor
#5, sample taken at 4uS, sample duration of 3.5uS


Picture #6, scope set at 5 uS, full cycle view

Shielding only the top of the coil form, the coil is just lying loose in the form.

Tinkerer

[bbsailor]

Tinkerer,

Do you live near a West Marine store?

If so, stop by and pickup some sailboat cable covers. They come in a variety of sizes from 3/32" ID to 3/8" ID. They are all 6 ft long and slit lengthwise to allow them to be slipped over sailboat rigging wire. These cable covers also make a good coil spacer that can be easily shielded and fit tightly inside a coil housing.

Let me know if you need more details about how to use these to make a PI coil?

bbsailor