PI limit to early sampling

[Ferric Toes]

Sampling down to 0 requires a very wideband null on the coil. I use lower inductances, I think the last time I did it, TX=100u and RX=250-300uH. Use a bit of ferrite for the X null, a bit of foil for the R null. I also run a low preamp gain, typically 100.

Hi Carl,
I remember you gave me some AD8055 IC's some years back. They made a super fast wide band preamp with no spurious responses. The trade off was that they were noisy. I got around this in one design by having a low noise preamp with low gain in front of the AD8055 and that worked better. I think I had a gain of 10 for the first amp (which also maximised its bandwidth) then followed with x50 for the 8055.

Eric.

[KingJL]

Hi Carl,
I remember you gave me some AD8055 IC's some years back. They made a super fast wide band preamp with no spurious responses. The trade off was that they were noisy. I got around this in one design by having a low noise preamp with low gain in front of the AD8055 and that worked better. I think I had a gain of 10 for the first amp (which also maximised its bandwidth) then followed with x50 for the 8055.

Eric.

On my first PI build, one of Carl's smd Hammerheads, I used two stacked boards and some minor jumpering to implement a 2 stage pre-amp and 2 channel receiver/detector. In it I used the AD797 @ 10x as the first stage pre-amp and an AD8055 @ 100x as the second stage pre-amp. If I remember right I had to add a 2pf cap across the feedback resistor of the AD797 as it wanted to supply it's own signal at such a low gain. Was very happy with the final result. The only drawback was the current consumption of the AD797. Solved that by populating the second board (remember, I had 2) power supply.

[green]

I found that the damping on the TX coil was critical at short delays. Try a fixed resistor in series with a reasonable wattage ceramic trimmer and adjust for the flattest RX response with quickest recovery. It looks in your last .png that there is some undershoot in the amplified RX and when you introduce a target, it is cancelling the undershoot before looking like a sensible signal. Try a damping preset trimmer and see if that gets rid of the undershoot.

Eric.

Tried adjusting damping. Adjusted damping for minimum Tx coil decay undershoot. Adjusting a ferrite bead to trim IB did better at trimming the amplifier out undershoot. Results still the same, signal started at zero instead of undershoot. Still needed to put copper coin and ground over opposite figure8 Rx coils to get same response. May be able to sample closer to zero delay time with an IB coil but I think you see the targets X signal until the Tx current has decayed making it harder to eliminate the ground signal. Probably have to much preamp gain and coil inductance for a good test near zero delay time. Getting over 1 million amps/sec Tx coil decay could be some of the reason the amplifier saturates. Just my thoughts, maybe not correct.

[Carl-NC]

Eric's post reminded me, damping is important and I recall I had to dampen the RX and well as the TX coil. On a figure-8 RX, you will need to dampen each half individually.

Also, when I run a gain of 100, it's using 2 stages of 10 each. Generally I use a little bit of cap on the feedback to limit noise, 0.5-1us TC.

Finally, yes, as long as there is any TX current decay there will be a reactive signal. When I say "0us" that starts from the point where the TX current is fully off.

[green]

Eric's post reminded me, damping is important and I recall I had to dampen the RX and well as the TX coil. On a figure-8 RX, you will need to dampen each half individually.

Also, when I run a gain of 100, it's using 2 stages of 10 each. Generally I use a little bit of cap on the feedback to limit noise, 0.5-1us TC.

Finally, yes, as long as there is any TX current decay there will be a reactive signal. When I say "0us" that starts from the point where the TX current is fully off.

Thanks, A difference where 0usec starts, I've been using the coil fet driver command turning off to external trigger the scope(0usec). Do you have an approximate number of coil TC's (L/Rd) that you would consider the Tx current being fully off?

[Ferric Toes]

Thanks, A difference where 0usec starts, I've been using the coil fet driver command turning off to external trigger the scope(0usec). Do you have an approximate number of coil TC's (L/Rd) that you would consider the Tx current being fully off?

That is what I use, but there is quite a bit of time which is 'dead' before you can sample. The Mosfet needs a low impedance driver both to turn on and turn off otherwise you have a TC due to the gate capacitance. If the Mosfet avalanches on the high voltage flyback, that eats up a few uS; then you have the L/R TC due to the damping resistor and finally the recovery time of the preamp stage. If you have a gate that only switches the preamp to the coil when all the foregoing transients are gone, then that overcomes the saturation recovery of the preamp.

My favourite solution to short <10uS delays is to have less pulse current (<0.5A) but put up the repetition rate. This helps also, with suitable selection of Mosfet, to avoid the avalanche situation so that the coil decay is controlled only by the damping resistor.

Eric.

[bbsailor]

That is what I use, but there is quite a bit of time which is 'dead' before you can sample. The Mosfet needs a low impedance driver both to turn on and turn off otherwise you have a TC due to the gate capacitance. If the Mosfet avalanches on the high voltage flyback, that eats up a few uS; then you have the L/R TC due to the damping resistor and finally the recovery time of the preamp stage. If you have a gate that only switches the preamp to the coil when all the foregoing transients are gone, then that overcomes the saturation recovery of the preamp.

My favourite solution to short <10uS delays is to have less pulse current (<0.5A) but put up the repetition rate. This helps also, with suitable selection of Mosfet, to avoid the avalanche situation so that the coil decay is controlled only by the damping resistor.

Eric.

Eric,

When you raise the TX frequency to about 3K PPS and then integrate the RX pulses, you improve the signal to noise ratio as you are detecting a signal synchronized to the TX pulse. How long the target stays in the coil area determines how many signals are sampled and integrated. Coil size and sweep speeds then become an important consideration. My early research on this after you revealed your integration technique, many years ago, caused me to discover how this is similar to how lock-in amplifiers work where they seek to extract very weak signals in the presence of noise. Another thing I found is that for each 100pf capacitance I can eliminate in the coil or TX circuit, I can reduce the delay about 1 uS. Then the damping resistor can be a higher value thus speeding up the coil turn off or discharge time constant to better stimulate smaller targets.

Thanks for stimulating my interest in Pulse Induction machines.

Joseph J. Rogowski

[green]

Thanks, A difference where 0usec starts, I've been using the coil fet driver command turning off to external trigger the scope(0usec). Do you have an approximate number of coil TC's (L/Rd) that you would consider the Tx current being fully off?

Looked at amplifier out on a scope this morning with different targets including some LF ferrite bead cores which should only show a signal when Tx current is changing. Bigfoot style coil(Rx two 8 inch round) Tx(oval coil surrounding Rx). Looks like 10 TC's(L/R) might be a good number for a critical damped coil. 300uH/300R=1usec*10(10usec delay), 378uH/1000R=.378usec*10(3.78usec delay). Does 10 TC's sound way off? Calculated Rd for a 300uH coil with 100, 200 300p capacitance. 10 TC's would reduce the delay by about 1usec stated by bbsailor in reply #37.

[Carl-NC]

Thanks, A difference where 0usec starts, I've been using the coil fet driver command turning off to external trigger the scope(0usec). Do you have an approximate number of coil TC's (L/Rd) that you would consider the Tx current being fully off?

I shouldn't say 'fully' off, but substantially off. The high-slew portion of turn-off is where you're going to stress the balance of an IB coil, after 2-2.5 taus the slew rate is lower and you can usually see a good RX signal at the preamp, assuming the preamp didn't get slammed and has to come out of saturation.

I did all this for a project at White's which I abandoned when I departed, so I don't even have a waveform I can show you. My recollection is that I got it to a point where the preamp waveform never railed out, even during the high-slew turn-off. Obviously high mineralization will ruin this and swamp the preamp for several us, but it had value in that i could see target responses all the way down to zero.

[Ferric Toes]

Eric,
When you raise the TX frequency to about 3K PPS and then integrate the RX pulses, you improve the signal to noise ratio as you are detecting a signal synchronized to the TX pulse. How long the target stays in the coil area determines how many signals are sampled and integrated. Coil size and sweep speeds then become an important consideration. My early research on this after you revealed your integration technique, many years ago, caused me to discover how this is similar to how lock-in amplifiers work where they seek to extract very weak signals in the presence of noise. Another thing I found is that for each 100pf capacitance I can eliminate in the coil or TX circuit, I can reduce the delay about 1 uS. Then the damping resistor can be a higher value thus speeding up the coil turn off or discharge time constant to better stimulate smaller targets.

Thanks for stimulating my interest in Pulse Induction machines.

Joseph J. Rogowski

Hi Joe,
Your post prompted me to get out a development model of the Goldquest SS and run it up on a scope. This will run at a minimum delay of 8uS from the TXoff initiation. TX pulse is 30uS wide and 13K PPS. The coil is a tapped mono of 450uH total but with a tap at 150uH. The TX drives the tap and the RX connects to the whole coil. The TX Mosfet is a IRFD210 with 27R in series with the drain ON resistance, and with the R of the coil we have about 0.3A pulse current. The rise time is 8.8uS so the pulse is substantially flat top for over 20uS and at switch off there is no avalanche.

I like the idea of the tapped coil as a normal mono is constrained by the TX characteristics and is sub-optimum for the RX.

Using high pulse rates speeds up the response time considerably, so it is possible to increase the integrator TC somewhat to further improve the S/N and still sweep at 1m/sec.

The original GQ SS ceased production about 5 years ago but that used a standard 300uH mono and the minimum delay was 10uS with 10K PPS.

Eric.

[bbsailor]

Looked at amplifier out on a scope this morning with different targets including some LF ferrite bead cores which should only show a signal when Tx current is changing. Bigfoot style coil(Rx two 8 inch round) Tx(oval coil surrounding Rx). Looks like 10 TC's(L/R) might be a good number for a critical damped coil. 300uH/300R=1usec*10(10usec delay), 378uH/1000R=.378usec*10(3.78usec delay). Does 10 TC's sound way off? Calculated Rd for a 300uH coil with 100, 200 300p capacitance. 10 TC's would reduce the delay by about 1usec stated by bbsailor in reply #37.

Green,

The mono coil discharge TC has two stages since the input resistor to the first amplifier stage is typical 1K ohm attaches to the two clamping diodes that essentially puts the Rd value in parallel with this IK ohm input resistor. Lets assume for ease of creating a good mental picture that Rd is 1K ohm and Rin is 1K ohm.

Stage 1: The coil discharge TC while the input voltage is over 0.6 V (diode conducting voltage) is the coil resistance divided by 500 ohms (combined value of Rd and Rin in parallel)
Stage 2: Then, when the voltage falls below 0.6V only the value of Rd affects the discharge TC.

This makes a coil discharge curve that has sort of a kink in it at about 0.6V.

The practical TC limit is 5 TCs to raise to 95.5 percent of maximum current or for a target to give up 95.5 percent of its eddy current signal. See this web link http://www.learnabout-electronics.or.../dc_ccts45.php

While 10TCs is the theory for current to rise to 100 percent, in practice 5 TCs is sufficient. A target that has a 3 uS TC will fall into the noise level in 5 TCs or 15 uS. The sooner you sample the higher the residual signal in the target will be. Small targets must be fully stimulated to have the highest potential to be detected. Here is where some reverse engineering can help optimize your coil, PPS frequency, TX and RX parameters to optimize detection of any particular target. A 3 uS TC target should have a coil discharge TC of 3/5 or 0.6 uS. A 300 uH coil with a 500 Ohm combined Rd will have a discharge TC of 0.6 uS down to 0.6V and then kink down with a TC of 0.3 uS.

If I remember correctly, Eric Foster reported that a target that receives a turn off current TC 5 times faster than it's own TC will respond just as well as if the turn off current occurred instantly. This is why those seeking low TC gold nuggets need to understand these rules to see if their equipment is optimum for their primary targets.

Beach hunters seeking coins or rings with longer TCs face another set of problems.

1. Coil size for good depth and area coverage.
2. Black sand, is it organic or is it magnetic.
3. Wet sand minimum delay versus dry sand minimum delay.
4. Mono versus DD coils. DD coils don't have Rin and Rd in parallel as they are two separate circuits and DD coils can operate a few uS faster than mono coils.

I hope this helps?

Joseph J. Rogowski

[bbsailor]

Green,

The mono coil discharge TC has two stages since the input resistor to the first amplifier stage is typical 1K ohm attaches to the two clamping diodes that essentially puts the Rd value in parallel with this IK ohm input resistor. Lets assume for ease of creating a good mental picture that Rd is 1K ohm and Rin is 1K ohm.

Stage 1: The coil discharge TC while the input voltage is over 0.6 V (diode conducting voltage) is the coil resistance divided by 500 ohms (combined value of Rd and Rin in parallel)
Stage 2: Then, when the voltage falls below 0.6V only the value of Rd affects the discharge TC.

This makes a coil discharge curve that has sort of a kink in it at about 0.6V.

The practical TC limit is 5 TCs to raise to 95.5 percent of maximum current or for a target to give up 95.5 percent of its eddy current signal. See this web link http://www.learnabout-electronics.or.../dc_ccts45.php

While 10TCs is the theory for current to rise to 100 percent, in practice 5 TCs is sufficient. A target that has a 3 uS TC will fall into the noise level in 5 TCs or 15 uS. The sooner you sample the higher the residual signal in the target will be. Small targets must be fully stimulated to have the highest potential to be detected. Here is where some reverse engineering can help optimize your coil, PPS frequency, TX and RX parameters to optimize detection of any particular target. A 3 uS TC target should have a coil discharge TC of 3/5 or 0.6 uS. A 300 uH coil with a 500 Ohm combined Rd will have a discharge TC of 0.6 uS down to 0.6V and then kink down with a TC of 0.3 uS.

If I remember correctly, Eric Foster reported that a target that receives a turn off current TC 5 times faster than it's own TC will respond just as well as if the turn off current occurred instantly. This is why those seeking low TC gold nuggets need to understand these rules to see if their equipment is optimum for their primary targets.

Beach hunters seeking coins or rings with longer TCs face another set of problems.

1. Coil size for good depth and area coverage.
2. Black sand, is it organic or is it magnetic.
3. Wet sand minimum delay versus dry sand minimum delay.
4. Mono versus DD coils. DD coils don't have Rin and Rd in parallel as they are two separate circuits and DD coils can operate a few uS faster than mono coils.

I hope this helps?

Joseph J. Rogowski

Pardon my fat fingers and slow brain. The actual current rises to 99.5 percent not the 95.5 percent that I stated below.

Sorry.

Joseph J. Rogowski

[green]

Hi Carl

I keep reading I should be able to sample down to 0usec. I'm using a Bigfoot style coil, two 8 inch round coils Rx(647uH, 1.05MHz SRF). Oval Tx surrounding Rx(378uH, 895kHz SRF)SRF measured with coil leads connected to coil. Differential input preamp(gain about 375). Tx, 160usec, constant rate, 6250 amps/sec, 1 amp peak. The coil is very close to balance without trimming with a ferrite bead core. Including some scope pictures, coil volts and preamp out volts. Top, ferrite near one end to unbalance coil when charging. Middle, ferrite near center to balance preamp out when coil is on. Bottom, Ferrite opposite end. I can zero preamp out when coil is on and change the waveform when Tx is turned off but I can't get rid of the saturated signal. Any suggestions what or where my problem is? Coil, preamp or something else.

I have an integrator that samples amplifier out during the last 137usec of the 160usec Tx. Was thinking using it to identify ferrous non ferrous but the ground signal is a major problem. Looking at it again to try and learn something. Charted some data. Coil off integrator(6usec delay, 10.2usec sample). The hot rock has the lowest coil off signal with almost the highest coil on signal. Is target shape causing some of the differences or is it mostly material? I thought it interesting to see the results of the different targets. It would be more interesting to know what causes the differences, comments appreciated. Targets were centered on one of the figure8 coils with a 15mm spacer between target and coil. The coins were added for reference signals. Charted integrator out(no target mv-target mv).

[Phiphi]

Some analog PIs use the difference between a slow and a fast integrator.

Do you have schematics of some of these analog PIs because I think it's a good idea ?
But I didn't see something like that in the schematics I have examined so far. Does it mean this idea was not good ?

[Phiphi]

As promised, I have performed some PSpice simulations with the aim to improve the behavior of PI Tx and Rx. First, I'll speak about Tx.
Below is the classic schematics with a critical damping resistor



And now the same one with a 47uH inductance inserted serially with the damping resistor which has been optimized to get the shortest decay:



Let's have a look to the Tx waveform after Tx off command on the Mosfet gate :



You can see that with the inductance the voltage peak is higher and the decay time lower. I look now what is the delay to reach very low voltage levels (100 uV) :



We have 4.9 us without inductance and 3.8 us with so a 1.1 us gain !
I will continue in the next post.