GEB

[green]

Thought I would record some US coin decays and see if I could use Excel to experiment with delay and sample timings. I'm using Tx=160usec, target delay=6usec, target sample=10.2usec, ground delay=4usec, ground sample=100usec. with my detector. Set record time for over 200usec to try longer sample times. Didn't work as good as I would like. Not enough amplitude or time resolution with my scope recording. Had to reduce target sample time from 10.2usec to 8.8usec to cancel ground signal. Not what I was hoping but maybe someone can learn some thing from the data. A picture is worth a 1000 words. I've been trying to chart the data so when looking at the chart I could get a feeling for signal increase or decrease with GB on. The log-log chart seems to be best(straight line decay for ground), but still not easy to tell how much signal increases or decreases. If the target decay is parallel to ground decay the signal would be zero. Any ideas on reading the chart or a different chart would be appreciated. I can experiment with timings on the bench but was hoping the chart would show why the GB signal was getting better or worse for different targets.

Looks like I recorded the GB ground - when it should have been +.

[mohalothman]

nice job

[green]

Got my Excel PI simulation program where it is easier to use. Charted PI hole with timings and ground slope(-1.35). About what I get with 160us constant rate TX. Simulation 1000 pulses/second. Changed delay time 6, 10, and 14us delay and adjusted target sample to cancel ground. Charted (GEB on reading/GEB off reading*1000). Wondering if using constant current Tx(ground slope closer to -1)or different timings would be better. I have posted the simulation before, if anyone wants the updated program I can post it. Any thoughts?

[green]

Trying to find best way to chart. (GEB hole_5)input the same for all charts. Top row: first delay 6, 10 and 14us, GEB sample time 100us, changed target sample time to cancel ground. Bottom row: first delay 6, 10 and 14us, target sample time 10.2us, changed GEB sample time to cancel ground. Top row,changing target sample time to cancel ground: Rx amplitude higher, larger TC shift in hole(better if using to different delay times to eliminate hole). 6 and 10us delay or 10 and 14us delay not bad. Still not sure the best way to chart?

[Ferric Toes]

Got my Excel PI simulation program where it is easier to use. Charted PI hole with timings and ground slope(-1.35). About what I get with 160us constant rate TX. Simulation 1000 pulses/second. Changed delay time 6, 10, and 14us delay and adjusted target sample to cancel ground. Charted (GEB on reading/GEB off reading*1000). Wondering if using constant current Tx(ground slope closer to -1)or different timings would be better. I have posted the simulation before, if anyone wants the updated program I can post it. Any thoughts?

Very interesting plots and I have not seen this done before. The shape of the notch is what I had expected and I wonder if it can be improved on to reduce the drop in detection range for targets either side of the maximum attenuation. How about a 'brick wall' bandpass filter that can be varied in width. This could then be adjusted to cope with situations where stones and rocks with slightly different decay rates to the background ironstone existed, as Kingswood found in parts of Western Australia. In areas were there was little or no variation in ground effect, the notch could be narrowed down. Probably need to go digital to do this.

Eric.

[green]

Trying to find best way to chart. (GEB hole_5)input the same for all charts. Top row: first delay 6, 10 and 14us, GEB sample time 100us, changed target sample time to cancel ground. Bottom row: first delay 6, 10 and 14us, target sample time 10.2us, changed GEB sample time to cancel ground. Top row,changing target sample time to cancel ground: Rx amplitude higher, larger TC shift in hole(better if using to different delay times to eliminate hole). 6 and 10us delay or 10 and 14us delay not bad. Still not sure the best way to chart?

Some thoughts. GEB hole_6 shows GEB lose better than GEB_5. GEB_5: target start amplitude the same at 3.9usec after Tx off. Thinking _5 would show effect sample time has on short TC targets. Shows lose in signal but short TC same size targets would start at higher amplitude than long TC targets. Charting(GEB on reading/GEB off reading*1000)maybe the best way? For GEB, making first delay adjustable(minimum to ?us)then adjusting target sample time, second delay time or GEB sample time to cancel ground signal makes the most sense? Which is better,adjusting target sample time, second delay time or GEB sample time? Any other thoughts?

I wonder if it can be improved on to reduce the drop in detection range for targets either side of the maximum attenuation.(Eric reply #5) Wondering the same.

[waltr]

Just some thoughts:

1- with set sample and delay times there is one setting of the GEB to null soil (ground).
This also creates a Null, "GB Hole", at some target TC.

2- With different set sample and delay times the "GB Hole" Nulls at different target TC.

3- With processor control of the sample and delay timing- have two sets of timing to shift the "GB Hole".
Soil is still nulled but Target become visible.

[Carl-NC]

Just some thoughts:

1- with set sample and delay times there is one setting of the GEB to null soil (ground).
This also creates a Null, "GB Hole", at some target TC.

2- With different set sample and delay times the "GB Hole" Nulls at different target TC.

3- With processor control of the sample and delay timing- have two sets of timing to shift the "GB Hole".
Soil is still nulled but Target become visible.

I built this a few years ago but in analog. A 4-channel PI where 2 channels create Hole1 and the other 2 channels create Hole2. Then I combined the results for no overall hole. This also results in 3 tones if you want: below Hole1 = high tone; between Hole1 & Hole2 = mid tone; above Hole2 = low tone.

[green]

Another variable. Constant rate Tx or constant current Tx. I get a different decay slope for ground if Tx time or shape changes. A US quarter has a near straight line decay on a log-log chart the first100us. The quarter's slope also changes with Tx time and shape. Maybe Eric, after his move could confirm slope change or anyone else. Would cr Tx or cc Tx give better S/N if using GEB(same average Tx current)? The chart was done with same average current for cr or cc with same on times. Does the ground decay slope have any effect on the GEB hole shape?

[Mdtoday]

Got my Excel PI simulation program where it is easier to use. Charted PI hole with timings and ground slope(-1.35). About what I get with 160us constant rate TX. Simulation 1000 pulses/second. Changed delay time 6, 10, and 14us delay and adjusted target sample to cancel ground. Charted (GEB on reading/GEB off reading*1000). Wondering if using constant current Tx(ground slope closer to -1)or different timings would be better. I have posted the simulation before, if anyone wants the updated program I can post it. Any thoughts?

Hi Green,
Nice work thanks for sharing and interesting results.
Would you mind posting your Excel file, I would be interested in trying it.

Cheers

mdtoday

[green]

Hi Green,
Nice work thanks for sharing and interesting results.
Would you mind posting your Excel file, I would be interested in trying it.

Cheers

mdtoday

Still trying to make it better. Lot easier to use now after the last update.

Any questions or improvements let me know.

[6666]

Good work.

[Mdtoday]

Still trying to make it better. Lot easier to use now after the last update.

Any questions or improvements let me know.

Great work and thank you fir posting, much appreciated

cheers

mdtoday

[green]

Another chart. Charts GEB as % of no GEB. Picture shows addition to Excel. Maybe a good way to show improvements instead of attaching the whole program(lot less memory).

I see P3 and N3 should have been one cell to the right in the picture.
I need to wakeup before posting, forgot the 3 in O3 in the formula.

[Ferric Toes]

Another variable. Constant rate Tx or constant current Tx. I get a different decay slope for ground if Tx time or shape changes. A US quarter has a near straight line decay on a log-log chart the first100us. The quarter's slope also changes with Tx time and shape. Maybe Eric, after his move could confirm slope change or anyone else. Would cr Tx or cc Tx give better S/N if using GEB(same average Tx current)? The chart was done with same average current for cr or cc with same on times. Does the ground decay slope have any effect on the GEB hole shape?

Yes, I can confirm that the US quarter has a problem within the first 100uS and changes with TX on-time. This is because the high conductivity of a quarter, both clad and silver, requires a TX width of at least 450uS for full energisation. I can see this clearly on my tester as I increase the TX on-time from 90uS minimum up to 450uS. Above 450uS there is no measurable change in shape. My system is constant current only, so that removes another variable as in constant rate. I will plot some mineralised soil on a log/log graph and superimpose the plot for a quarter that should curve increasingly downward from the linear plot for the soil. We need to go to longer cycle time to do this though, maybe up to 1mS. If we design a detector to give the best range on higher conductivity targets then changes in the timing of everything needs to be addressed, including GB samples and delays. More later.

Eric.