Chance PI coil...

[baum7154]

Did you tried this coil design, and if you did, what is wrong with it? I mentioned this as interesting design approach, maybe not specifically for Chance. Also, please you read this topic, this is all about making different coil type for Chance, in order to improve performance, because “original” coil turns out to be most problematic part of this detector, limiting it's usability.

What is wrong with this coil? This coil is constructed in 4 layers of 8 turns per layer. Because of this, one face of the coil is tied to the HOT TX drive conductor and the other face is tied to the GROUND conductor. This necessitates the application of a graphite shield in order to operate at short delays but this shield also adds more capacitance and slows it down. That is fine as most coils fit this profile. For CHANCE, this coil is a departure from the coil design parameters that the detector was optimized to accommodate. To improve the construction of this coil I would put an inner circle barrier form and and outer circle barrier form on the coil ID & OD before it gets flooded with the urethane foam. This would give a clean profile on three sides of the coil leaving only the top to be sculpted to shape.

The reason I prefer the #1 coil on the toroidal plastic form detailed in this thread is that the outside windings are at ground and surround the inner windings, the innermost of which are tied to the coil drive lead. It is harder to construct but worth the trouble in my opinion. This coil is also a departure from the original CHANCE parameters and the penalty for that is unclear. What is clear is that the original CHANCE coil has not performed for my requirement of detecting small gold but it is Ok for coins, some jewelry, and other artifacts. It would help a lot if we had the source code for CHANCE to help optimize it for gold. That said I am sure my current CHANCE PI with spidercoil does detect small gold as it is today.

Dan

[Tepco]

My question was intended for KT315, otherwise I agree with your comment. Actually, even more modifications may be advisable, different layer forming, eventually omitting layer1, different turn number etc, his is just raw idea about mechanical shape, some experiment with this shape can be interesting.

[baum7154]

My question was intended for KT315, otherwise I agree with your comment. Actually, even more modifications may be advisable, different layer forming, eventually omitting layer1, different turn number etc, his is just raw idea about mechanical shape, some experiment with this shape can be interesting.

Yes it definitely has the benefit of not needing a separate plastic coil former and that makes construction much easier.

[baum7154]

[QUOTE=baum7154;184199]

First of all I have not tried to detect a 1/4" X 1/4" target of either foil or aluminum can. 1/4" X1/4" is 1/16 of a square inch. I have detected a 1/4 square inch target of aluminum can and it is a great target for testing, measuring 1/2" X 1/2". Perhaps the language I used caused some confusion. Sorry if it did. If you are detecting targets 1/4" X 1/4" of aluminum can that is impressive. My most difficult standardized target so far is the 1/2" X 1/2' .001" thick aluminum foil target.

Just performed detection distance on a 1/4" X 1/4" piece of .004" thick aluminum can and my CHANCE PI with #1 coil and 1100 Rd sees the sample at 2.5" near coil center. It does not see a 1/4" X 1/4" piece of .001" thick aluminum foil probably due to being too short of a decay for it. I used a 1/2" X 1/4" piece of .001 thick foil and folded it to a 1/4" X 1/4" square .002" thick and it is seen at 1.5".

Dan

To the group:

I'm trying to deduce how many microseconds the CHANCE PI is sampling when the Guard Interval is set to the minimum of 10. At this setting it sees the 1/4 X 1/4"sample of aluminum can at 2.5". Per Green's plot in this thread this sample has a TC of 1.5us. The detector also sees (@1.5") a 1/4" X 1/4" piece of foil with a total thickness of .002 or half as thick as the aluminum can. From that can I expect the TC of this sample to be about .75 us? With a total decay of 5 X the TC that puts the first sample at 7.5us and the second sample at 3.75us. Is this correct? If so then is the detector effectively operating it's sample & hold in the neighborhood of about 3.75us? I appreciate your help and input.

Dan

[Tinkerer]

[QUOTE=baum7154;184293]

To the group:

I'm trying to deduce how many microseconds the CHANCE PI is sampling when the Guard Interval is set to the minimum of 10. At this setting it sees the 1/4 X 1/4"sample of aluminum can at 2.5". Per Green's plot in this thread this sample has a TC of 1.5us. The detector also sees (@1.5") a 1/4" X 1/4" piece of foil with a total thickness of .002 or half as thick as the aluminum can. From that can I expect the TC of this sample to be about .75 us? With a total decay of 5 X the TC that puts the first sample at 7.5us and the second sample at 3.75us. Is this correct? If so then is the detector effectively operating it's sample & hold in the neighborhood of about 3.75us? I appreciate your help and input.

Dan

I suppose you do not have an oscilloscope to look at the sample pulses? The samples have a certain width, maybe 5us or 10us or so. The sample is the average of the signal during this time.
If there is a feedback capacitor on the pre-amp, often something like 2 to 5pf, this stretches or delays the peak target response so a very short TC target can be measured a little longer. I would need to see the schematics.
Anyway, your PI is very sensitive to small targets, mostly due to your well built coils.

One question, I believe the ChANCE PI is supposed to discriminate iron. Have you made any tests? Your very fast damping may possibly affect the iron discrimination.

[baum7154]

[QUOTE=Tinkerer;184295]

I suppose you do not have an oscilloscope to look at the sample pulses? The samples have a certain width, maybe 5us or 10us or so. The sample is the average of the signal during this time.
If there is a feedback capacitor on the pre-amp, often something like 2 to 5pf, this stretches or delays the peak target response so a very short TC target can be measured a little longer. I would need to see the schematics.
Anyway, your PI is very sensitive to small targets, mostly due to your well built coils.

One question, I believe the ChANCE PI is supposed to discriminate iron. Have you made any tests? Your very fast damping may possibly affect the iron discrimination.

_____________________
I do have a Tektronix 7603 scope with a 7B53A timebase plug in. I have been too busy to get the scope on it but this is high on my list to document Guard Interval vs microseconds as I have never seen it anywhere. It just occurred to me after testing the 1/4" X 1/4" Al samples that they are pretty short decays and when CHANCE saw them it made me wonder just how fast the detector was sampling. Does my original assessment on sample times make any sense?

There is no capacitor on the OP37 input amp so only parasitic capacitances from the circuitry are there.

Yes CHANCE does discriminate for metal types and I have been playing with Rd values to try to optimize sensitivity without degrading discrimination. Right now I am at 1040 Ohms and this seems to be a pretty good compromise. There is some jumping of the VDI indication from Right to Left (Left side is idle default) with a 7/16" iron hex head from a 1/4-20 bolt face on to the coil, but on edge it seems to lock on at the Iron end of the scale. A new .32 cal lead projectile wants to display most of the time as gold , jumping to it's proper position about 30% of the time. Copper penny and Silver dime are rock solid where they should be on the VDI. 10.8 grain piece of dental gold is about 90% on the left or Gold end. And yes , if I push Rd too high, in the range of 1200 ohms and above then everything tends do want to display on the left /default side of the VDI as Gold/Al. I have not tried more testing at higher GI settings to see if discrimination changes because I am building for small gold detection.

I am thinking of putting a 10 position rotary switch and resistor values to set Rd within a limited range near the cable connector to allow adjustment of Rd from about 950 ohms to about 1120 ohms. The idea is to tune the Rd for best response in different soil conditions. The way it is now with setting Rd in air I may end up significantly underdamped when the detector is in use over different soils, compromising detector performance.

Dan

[green]

The 12 bit A-D (MCP3201) reads the preamp output. Have you tried looking at amplifier input and pin 5 of the A-D while adjusting the Guard Interval?

[baum7154]

No I'll try to do that tonight.

[green]

I was trying to simulate a target with a 1.5usec time constant. Including a plot. Might be a start. I was wondering what effect amplifier Fc has. Your coils have a SRF about 1Mhz. I calculate a Fc of about 150 khz for the Chance amplifier. I'm wondering why a critical damped coil around 1Mhz has anything to do with the response. The simulation plot doesn't include saturation from fly back. Haven't figured how to include it. Looks like a lower amplifier Fc gives a higher voltage at longer delays if saturation doesn't cause a problem. To get speed my amplifiers have a Fc>> 1Mhz. Probably a mistake. The simulation is a 10 mv step which should cause a 160 mv peak at amplifier out with a high amplifier Fc.

[baum7154]

The 12 bit A-D (MCP3201) reads the preamp output. Have you tried looking at amplifier input and pin 5 of the A-D while adjusting the Guard Interval?

I looked at the A-D pin 5 on Channel 1 and the diodes at the op amp input on Channel 2 also used it for triggering. The TX signal is very complex with 19 short pulses of slightly increasing duration followed by a long pulse and flyback after that. After much examination of several groups of A-D pulses I did find the one pulse that moved with changes to the Guard Interval. At a GI setting of 10 the pulse followed 2.5 us after flyback return to zero and the pulse was 12.5 us wide. Raising the GI to 20 the pulse started 12.5 us after RTZ. The pattern appeared to be almost 1 GI for 1 us but as the GI inputs increased I did notice that with 80 units increase of GI the A-D pulse moved only 71us. (Maybe my scope is not what it once was.) Anyway my earlier question about sampling in the neighborhood of 3.75us appears to be answered as the pulse command starts at about 2.5 us and is 12.5 us wide with a GI =10. For all practical purposes 10 GI units increase = 10 us increase and GI =10 =2.5us or so.

[Tinkerer]

I was trying to simulate a target with a 1.5usec time constant. Including a plot. Might be a start. I was wondering what effect amplifier Fc has. Your coils have a SRF about 1Mhz. I calculate a Fc of about 150 khz for the Chance amplifier. I'm wondering why a critical damped coil around 1Mhz has anything to do with the response. The simulation plot doesn't include saturation from fly back. Haven't figured how to include it. Looks like a lower amplifier Fc gives a higher voltage at longer delays if saturation doesn't cause a problem. To get speed my amplifiers have a Fc>> 1Mhz. Probably a mistake. The simulation is a 10 mv step which should cause a 160 mv peak at amplifier out with a high amplifier Fc.

Even with a SFR of 1MHz, there are still the eddy currents within the coil wire to be accounted for.

There is also another way to look at it. We can get resonance at a certain SFR in a coil with high inductance and low capacitance or the same SFR with high capacitance and low inductance. However, with a traditional PI, we know that the stray capacitance plays an important role in the damping. If we look at the first few ns of the switch OFF transient, we can clearly see what the capacitance does.

I have a slow coil and a TX and preamp circuit on the breadboard at the moment, where I can observe the target response at time 0. According to what I see, for these short TC targets, we need at least 200 kHz, 1MHz is excellent but maybe a bit overkill. Once I have a fast coil I will probably know more.


Could you please zip the .asc file of your simulation and post it?

[Tinkerer]

CHANCE guard interval.

I looked at the A-D pin 5 on Channel 1 and the diodes at the op amp input on Channel 2 also used it for triggering. The TX signal is very complex with 19 short pulses of slightly increasing duration followed by a long pulse and flyback after that. After much examination of several groups of A-D pulses I did find the one pulse that moved with changes to the Guard Interval. At a GI setting of 10 the pulse followed 2.5 us after flyback return to zero and the pulse was 12.5 us wide. Raising the GI to 20 the pulse started 12.5 us after RTZ. The pattern appeared to be almost 1 GI for 1 us but as the GI inputs increased I did notice that with 80 units increase of GI the A-D pulse moved only 71us. (Maybe my scope is not what it once was.) Anyway my earlier question about sampling in the neighborhood of 3.75us appears to be answered as the pulse command starts at about 2.5 us and is 12.5 us wide with a GI =10. For all practical purposes 10 GI units increase = 10 us increase and GI =10 =2.5us or so.

If the preamp has a Fc of 150kHz, this acts as an integrator for the very short TC signal response. Then the 12.5 us wide sample further integrates the signal. Less integration would make a higher peak response signal but it would be noisier.

[green]

I looked at the A-D pin 5 on Channel 1 and the diodes at the op amp input on Channel 2 also used it for triggering. The TX signal is very complex with 19 short pulses of slightly increasing duration followed by a long pulse and flyback after that. After much examination of several groups of A-D pulses I did find the one pulse that moved with changes to the Guard Interval. At a GI setting of 10 the pulse followed 2.5 us after flyback return to zero and the pulse was 12.5 us wide. Raising the GI to 20 the pulse started 12.5 us after RTZ. The pattern appeared to be almost 1 GI for 1 us but as the GI inputs increased I did notice that with 80 units increase of GI the A-D pulse moved only 71us. (Maybe my scope is not what it once was.) Anyway my earlier question about sampling in the neighborhood of 3.75us appears to be answered as the pulse command starts at about 2.5 us and is 12.5 us wide with a GI =10. For all practical purposes 10 GI units increase = 10 us increase and GI =10 =2.5us or so.

The A-D takes one reading each time pin 5 goes low. I think the time should be referenced the start of fly back, since the micro doesn't know what the fly back looks kike. It might be interesting to look at amplifier out (A-D in) and A-D pin 5.

[Could you please zip the .asc file of your simulation and post it?]
I'm computer illiterate, will try to learn how. I have 18 weeks of electronic training in the Army in 1959. Have designed and built some circuits with analog and logic chips. So far this has been another learning experience.

[baum7154]

The A-D takes one reading each time pin 5 goes low. I think the time should be referenced the start of fly back, since the micro doesn't know what the fly back looks kike. It might be interesting to look at amplifier out (A-D in) and A-D pin 5.

[Could you please zip the .asc file of your simulation and post it?]
I'm computer illiterate, will try to learn how. I have 18 weeks of electronic training in the Army in 1959. Have designed and built some circuits with analog and logic chips. So far this has been another learning experience.

I agree that the start of flyback is a know data point to the uc and it makes much better sense to use it. I'll get a measurement of how microseconds after start of flyback the 12.5 us window goes open.

[Tepco]

This is right way to measure, not op-amp input diodes, you can actually use one of MOSFET gate drive signals to synchronize scope. Then exact delay from TX end to ADC sample can be seen, certainly not 2.5uS, most of flyback time must be added to this value.