My take on the HH2

[waltr]

Thanks for the improved explanation.

I did once upon a time fiddle about with GB using variable pulse widths and only one EF, but it was tricky to implement, and the results were not that good.

Ok, that is about what I am finding.
Since this theory has been posted and discussed in other threads here I am willing to give it a try.

[green]

Ok, that is about what I am finding.
Since this theory has been posted and discussed in other threads here I am willing to give it a try.

I still think your three sample is the same as four sample when adjusted to cancel ground and you would loose a lot of distance with a quarter with GB adjusted to zero. Wondering if you could check detection distance with a US clad quarter with GB at zero and adjusted to cancel your red brick. Charted some ground and a quarter. If we could calculate average volts during sample time we could approximate required GB sample time. Should be able to approximate detection distance for the quarter with GB at zero using the quarter chart. Used 100usec constant current Tx when charting.

[waltr]

I have measured the Voltage out of the Integrator, Normalized to No Target Voltage, for different targets and GEB settings (these are posted above in this thread). This should be proportional to distance.

A US Quarter with GEB set to Zero: Mono coil = 0.59V, Concentric coil = 1.81V
A US Quarter with GEB set to 50us: Mono coil = -0.9V, Concentric coil = -2.6V

A US Dime with GEB set to Zero: Mono coil = 0.48V, Concentric coil = 1.21V
A US Dime with GEB set to 50us: Mono coil = -0.4V, Concentric coil = -0.22V

A Red brick gives NO change of the integrator output Voltage at any GEB setting.
So the brick I have may not contain much mineralization.
This is partly why I am unsure if this GEB scheme does work.

[6666]

I'll try a four sample method

Very interested in your results,

[Carl-NC]

If GEB = 50us then S1 = 15us, S2 = 65us, S3 = 50us.
IF GEB = 0 then S1 = 15us, S2 = 15us, S3 = 0us.
This seems correct?
Question is: is S2 at only 20us after S1 effective for EFE?
Or: does the three Sample method really not work and four samples are needed with the last two (more than 200us from the first two) widths be varied as GEB adjust?

This looks correct. I've never messed with the 3 sample method, but it should work. The reasons I can think why it might not are:
1. Switch charge injection is unbalanced; but I can't imagine this amounts to much.
2. Integrator droop is unbalanced; again, I tend to think not.

[Carl-NC]

I've said quite often, the best way to deal with EF is to use a bipolar transmitter. Then EF samples are not needed at all. All my PI designs are now bipolar using either an H-bridge (as in the case of the new Fisher pinpointer) or using a bifilar differential coil (a project I'm working on now).

I did once modify a White's TDI to use variable pulse width GB instead of variable gain. Pretty sure it still had a monopolar TX and the dual EF samples. It worked just fine, but I never did a detailed evaluation to see what improvements it offered. Seemed to perform about the same. Theoretically, GB using PW should offer a slight noise improvement over the gain method.

[green]

I have measured the Voltage out of the Integrator, Normalized to No Target Voltage, for different targets and GEB settings (these are posted above in this thread). This should be proportional to distance.

A US Quarter with GEB set to Zero: Mono coil = 0.59V, Concentric coil = 1.81V
A US Quarter with GEB set to 50us: Mono coil = -0.9V, Concentric coil = -2.6V

A US Dime with GEB set to Zero: Mono coil = 0.48V, Concentric coil = 1.21V
A US Dime with GEB set to 50us: Mono coil = -0.4V, Concentric coil = -0.22V

A Red brick gives NO change of the integrator output Voltage at any GEB setting.
So the brick I have may not contain much mineralization.
This is partly why I am unsure if this GEB scheme does work.

Thanks for the test. I was expecting a lower signal with GEB set to zero. Do you know what the first delay was?

[waltr]

Thanks for the test. I was expecting a lower signal with GEB set to zero. Do you know what the first delay was?

First delay: 10us with the Concentric coil, 12us with the Mono coil
Second delay: 20us.

Notice that with GEB at 50us the Voltage goes negative instead of positive.
This is what makes me think this is working.

[green]

First delay: 10us with the Concentric coil, 12us with the Mono coil
Second delay: 20us.

Notice that with GEB at 50us the Voltage goes negative instead of positive.
This is what makes me think this is working.

Added sample lines to charts. If you take the center of the samples for the quarter the second sample is about 1/2 the first so you loose about 1/2 the signal, maybe 1inch detection depth. I was thinking the average for the second sample would be closer to the first. I'm thinking your GEB sample would need to be around 100usec instead of 50usec to cancel ground.

[waltr]

Interesting.
I have been reading your threads on the log amp and TC measurements but didn't really understand what your graphs were showing.
Adding the Target and GEB samples to these graphs helps.

Will try running higher GEB sample times.

[waltr]

More mods

I thought it was a good idea to add gain after the sampling JFETs. WRONG, this has been causing issues in the SAT stage particularly with GEB sampling. When the GEB sample time is longer than 70usec then the SAT stage would saturate on the +/-5V rails. Also, at times there is an un-stability, SAT output would fluctuate wildly.
The post sampling op-amp is removed (U13 in schematic Sheet #2) and the resistors to ground (R83, R88, R81, R89) removed. Then jumpers soldered from JFET outputs (Q6 & Q7) to the Integrator input resistors (at TP8 & TP11).

Sensitivity/distance has not decreased but the Sat stage doesn't saturate or become unstable.
I had reduced the pre-amp gain originally to 500 (R13 is 2k not 1k). This did seem to help the NE5534 to recover faster. If more gain is desired then would be best to use a two stage front end like on many of the newer PI detector designs.

Now the GEB sampling time can be increased to 200usec without Sat stage issues.

[Tim]

Gain and 2 stage front end are both desirable waltr. I commend your efforts thus far. Sorry my limited experience cannot contribute more. I am champing at the bit to see a wholly successful schem and hex for board design contributions.

[waltr]

Added sample lines to charts. If you take the center of the samples for the quarter the second sample is about 1/2 the first so you loose about 1/2 the signal, maybe 1inch detection depth. I was thinking the average for the second sample would be closer to the first. I'm thinking your GEB sample would need to be around 100usec instead of 50usec to cancel ground.

Green,

Just re-read this thread:
http://www.geotech1.com/forums/showt...-Balance/page3
Did some calculations based on Davor's math and the sample/delay times I have been using.
Guess what? They match the GB sample time you show in the graphs.
With a 15us delay, 15us Target sample, then 20us delay2 the GB sample time should be 100usec.

this is based on Davor's equation in post 65 (see link):
Basically: Times, a = start of Target sample, b = end of target sample, c = start of GB sample, d = end of GB sample.
Then to balance ground: ln(b/a) - ln(d/c) = 0
Which is the same as: b/a - d/c = 0
So if a = 15, b = 30, c = 50, then d = 100us --as you predicted.

However, Davor's calc's doesn't seem to adjust to different grounds (different magnetic viscosity).

[green]

Green,

Just re-read this thread:
http://www.geotech1.com/forums/showt...-Balance/page3
Did some calculations based on Davor's math and the sample/delay times I have been using.
Guess what? They match the GB sample time you show in the graphs.
With a 15us delay, 15us Target sample, then 20us delay2 the GB sample time should be 100usec.

this is based on Davor's equation in post 65 (see link):
Basically: Times, a = start of Target sample, b = end of target sample, c = start of GB sample, d = end of GB sample.
Then to balance ground: ln(b/a) - ln(d/c) = 0
Which is the same as: b/a - d/c = 0
So if a = 15, b = 30, c = 50, then d = 100us --as you predicted.

However, Davor's calc's doesn't seem to adjust to different grounds (different magnetic viscosity).

I think Davor's equation predicts end of sample would be 100 which means ground sample is 50usec. Works if ground slope is -1 and EF sample is zero. With ground slope =-1.25 which I get with 100usec constant current Tx, ground sample needs to be around 100usec.

I have an Excel program that simulates a PI. The simulation predicts a EF reading. The 50usec ground sample would cancel a slope of -1 if EF sample=zero. The 100usec ground sample comes close to cancelling the -1.25 slope. Need to adjust it a little.

[green]

I think Davor's equation predicts end of sample would be 100 which means ground sample is 50usec. Works if ground slope is -1 and EF sample is zero. With ground slope =-1.25 which I get with 100usec constant current Tx, ground sample needs to be around 100usec.

I have an Excel program that simulates a PI. The simulation predicts a EF reading. The 50usec ground sample would cancel a slope of -1 if EF sample=zero. The 100usec ground sample comes close to cancelling the -1.25 slope. Need to adjust it a little.

if a = 15, b = 30, c = 50, then d =140. Excel simulator suggests 90usec ground sample with a = 15, b = 30, c = 50

I used a=12usec when I suggested near 100usec for GB