Perhaps I should have applied for a patent?

[KingJL]

Thanks, that is what I was hoping someone would do. Tried making the changes. Doesn't look like I could sample at 1usec. Maybe missed a change or something hidden(like coil capacitance). Could you post your spice model with changes?

Here it is. You may need to tweak the damping resistors to get to your 1 second sample. I did some tests a 2 usec and had a 30-40 uvolt change between "target" and "no target". I haven't refined it too much. I would try commenting the TX/RX K statemet and tuning the TX damp, the un-comment and tweak the RX damp. In Eric's real world setup, there seems to be no need for a TX damp. This could be, as he suggested, because of the properties of the shield fabric material (those properties that make it an effective emi shield).
Eric(mod).zip

[KingJL]

Here it is. You may need to tweak the damping resistors to get to your 1 second sample. I did some tests a 2 usec and had a 30-40 uvolt change between "target" and "no target". I haven't refined it too much. I would try commenting the TX/RX K statemet and tuning the TX damp, the un-comment and tweak the RX damp. In Eric's real world setup, there seems to be no need for a TX damp. This could be, as he suggested, because of the properties of the shield fabric material (those properties that make it an effective emi shield).
[ATTACH]41185[/ATTACH]

Try 27 for R1 and 580 for R2. Difference between "target and "no target" is (-)66 uV @ 1 usec.

[green]

Try 27 for R1 and 580 for R2. Difference between "target and "no target" is (-)66 uV @ 1 usec.

Thanks. Tried it but I'm missing something. Click on target and don't see target voltage. Click on Rx and don't see 5usec TC target decay, should be 11.5usec/decade. Getting late and I'm probably doing something wrong. Won't be able to look at again until afternoon 5 Dec.

[Idefix]

Hi,
I have made some others runs.
I have change the Mosfet, to 0.0xx Rdson , So the current , at 5V supplies is more than 10A. I had 8 previously.
I change the coupling model, so I use two transfo instead of one, with the 80 pf coupling in the middle.
And , a difference with LTspice runs , the coils are not connected , as it can be seen in light grey in my pictures.
I also have change the ref ground of the RX coil, adding R38 /1000 ohms.

About the target signal you use, I use a generator, but on your sims, I suppose Target is defined somewhere?
(I found on other thread :

* Simple example of how to simulate a target response in SPICE
* Residual RX signal
V1 residual 0 SIN 0 15m 10K
R1 residual 0 100MEG
* Strength of target signal (phase retarded by 60 degrees)
V2 strength 0 SIN 0 100m 10K 0 0 -60
R2 strength 0 100MEG
* Sweep signal
V3 sweep 0 PWL(0,0 10m,0 20m,1 80m,1 90m,0)
R4 sweep 0 100MEG
* Modulate strength signal with sweep signal to produce target response
E1 target 0 POLY(2) strength 0 sweep 0 0 0 0 0 1
R3 target 0 100MEG
* Combine residual and target signals to get RX signal
E2 rx_signal 0 POLY(2) residual 0 target 0 0 1 1
R5 rx_signal 0 100MEG
.tran 100u 100m
.plot v(residual) v(strength) v(sweep) v(target) v(rx_signal)
.end
)

Is that you uses?

Some pictures:

[green]

Thanks. Tried it but I'm missing something. Click on target and don't see target voltage. Click on Rx and don't see 5usec TC target decay, should be 11.5usec/decade. Getting late and I'm probably doing something wrong. Won't be able to look at again until afternoon 5 Dec.

Found what I was missing. Spice directive. The Rx signal is about .1 volt at 1usec. I have been thinking the no target decay would need to be closer to the target signal to detect. Just looking for a change so maybe not correct. With the no target signal at .1volt amplifier gain would be limited to around 50. Is there a no target signal level to try for when adjusting first delay?

It will be interesting to see what Eric ends up with and how close the spice models are to hardware.

[green]

Hi,
I have made some others runs.
I have change the Mosfet, to 0.0xx Rdson , So the current , at 5V supplies is more than 10A. I had 8 previously.
I change the coupling model, so I use two transfo instead of one, with the 80 pf coupling in the middle.
And , a difference with LTspice runs , the coils are not connected , as it can be seen in light grey in my pictures.
I also have change the ref ground of the RX coil, adding R38 /1000 ohms.

About the target signal you use, I use a generator, but on your sims, I suppose Target is defined somewhere?
(I found on other thread :

* Simple example of how to simulate a target response in SPICE
* Residual RX signal
V1 residual 0 SIN 0 15m 10K
R1 residual 0 100MEG
* Strength of target signal (phase retarded by 60 degrees)
V2 strength 0 SIN 0 100m 10K 0 0 -60
R2 strength 0 100MEG
* Sweep signal
V3 sweep 0 PWL(0,0 10m,0 20m,1 80m,1 90m,0)
R4 sweep 0 100MEG
* Modulate strength signal with sweep signal to produce target response
E1 target 0 POLY(2) strength 0 sweep 0 0 0 0 0 1
R3 target 0 100MEG
* Combine residual and target signals to get RX signal
E2 rx_signal 0 POLY(2) residual 0 target 0 0 1 1
R5 rx_signal 0 100MEG
.tran 100u 100m
.plot v(residual) v(strength) v(sweep) v(target) v(rx_signal)
.end
)

Is that you uses?

Some pictures:
[ATTACH]41188[/ATTACH][ATTACH]41189[/ATTACH][ATTACH]41190[/ATTACH]

For a PI, a thin or small target with little or no skin effect can be simulated with an inductor and resistor in parallel. Target time constant=L/R, KingJL example [Eric(mod).zip](50nH/.01ohms=5usec). Need to couple Tx to target and couple target to Rx, target to Rx not needed with a mono coil. Not sure about Tx to Rx coupling. Rx signal from target at Tx off typically 1uv to 1mv, can be higher or lower. Simulation trace looks very close to PI recorded trace.

[green]

I had a new idea that shows promise and that is using the wrap-around shield (with gap) as a TX on a RX coil. D=140mm, Lcoil(RX) = 325uH. Wrap-around shield 0.5uH and R = 0.4 ohms. TX pulse 30us, rise time approx 2uS then flat top at 10A. Back emf at switch off 12V. Only the RX coil appears to need damping. Pulse Rep rate is at present 3000pps and current drawn 500mA at 5V TX supply. This rep rate could be reduced to around 1000 so as to be more economical on current, but I am stuck with the 3K at present in order to get the 30uS pulse width.

TX switch is IRFP220N with on resistance of 0.075 ohms and VDss of 200V. It is on a heat sink but barely gets warm. Shield gets slightly warm. I coupled the RX winding to a NE5534 stage just to see what is happening and it looks interesting. Both switch on and switch off decays of a target are plainly seen and sampling could take place after 1uS. A faster front end will be tried next to get more detail.

Modelling the arrangement on Spice would no doubt give more insight if anyone has the time.

Eric.

The spice models are predicting a Tx back emf close to 200V at switch off instead of 12V. Any ideas why the difference, maybe the reason Rd not needed for Tx coil?

[Idefix]

Hi,
I have changed the coil modelling, by using coil with coef K . (And a target coil too ) . I cut coils in two half coils to have the 80p in the middle.
See screen copies.
The 10A current is reached , but the mosfet used oscillates without the resistor R7. 250v 38A but hight output capacitance 870pf . It generates others Auto TX 20 µs later, and 30µS..
The flyback voltage goes up to more than 50v, depending on the value or R7.
The coupling values needs to be looked closely, measured if possible. The behaviour changes with them.

[KingJL]

The spice models are predicting a Tx back emf close to 200V at switch off instead of 12V. Any ideas why the difference, maybe the reason Rd not needed for Tx coil?

I wouldn't worry too much about the inconsistency of the model vs the actual circuit as regards to the TX back emf and the damping of RX and TX. That is because I don't think the TX fabric characteristics can be easily modeled (if at all). I believe the reason for the difference of both the back emf and Rd(TX) is because of the high frequency absorption characteristics of the copper/nickel fabric (I think that would be extremely difficult to model). I had occasion to use similar materials back in my radar cloaking days. There are a LOT of interactions that you might normally not expect. Think of how the current in a conductor moves toward the surface of a conductor as the frequency of the change increases; then take a copper conductor and plate it with nickel (about 4 times the resistance); now have a bunch of kinks in the conductor with another conductor at right angle to it and in contact with it at each kink. There is a LOT going on there!!!

[Ferric Toes]

I wouldn't worry too much about the inconsistency of the model vs the actual circuit as regards to the TX back emf and the damping of RX and TX. That is because I don't think the TX fabric characteristics can be easily modeled (if at all). I believe the reason for the difference of both the back emf and Rd(TX) is because of the high frequency absorption characteristics of the copper/nickel fabric (I think that would be extremely difficult to model). I had occasion to use similar materials back in my radar cloaking days. There are a LOT of interactions that you might normally not expect. Think of how the current in a conductor moves toward the surface of a conductor as the frequency of the change increases; then take a copper conductor and plate it with nickel (about 4 times the resistance); now have a bunch of kinks in the conductor with another conductor at right angle to it and in contact with it at each kink. There is a LOT going on there!!!

Additionally, the tape is 1in wide and it is impossible to wind it on helically without small creases and folds. Each turn overlaps slightly resulting in double thickness at the overlap, so electrically I agree that it would be almost impossible to model. A small component of the collapsing field will interact with any crease or fold and absorb some of the energy, hence the apparent self damping.
I might get a reel of 1/2in tape as this will wind on smoother or make a larger RX winding; say 10 or 12in diameter for the 1in tape. One other advantage of the tape TX is that the large surface area helps to dissipate heat at high pulse currents.

Eric.

[Dutchie]

Dear Eric

As a newby reading this forum I have a few questions:
I do not fully understand why the turns of the shield should overlap. At a frequency of 100 MHz the wavelength is approx. 3 m.
Can we really expect high harmonics in the circuitry? Else working with a modest gap between the turns would make it a lot easier.
Has someone build the same coil and gotten te same results.
Can you please specify the materials you used and maybe some pictures of the hardware (coil + connecting circuitry) - I suppose the shield has at least 2 contacts with the circuitry?

Thanks for your time.

Karel

[Ferric Toes]

Hi Karel,

If the shield were just to minimise r.f. interference then a small gap would be fine. Also to prevent capacitive changes between the coil and conductive ground, also a small gap is OK. However, to use the shield as a single turn TX there must be an overlap, otherwise the shield will act as a toroidal coil around the RX coil with little external field. Of course, we are relying on the conductive adhesive in the overlap to complete the circuit between turns, but it seems to be effective.

The shield does have two contacts with the circuitry. One end is ground and the other end to the drain of the switching mosfet as in the Spice circuits in earlier posts. I haven't used a series diode though, but I gather that it appears not to make any difference.

Eric.

[KingJL]

Hi Karel,

"... I haven't used a series diode though, but I gather that it appears not to make any difference."

Eric.

Correct, except for a reduction of TX current!

[green]

Hi Karel,

If the shield were just to minimise r.f. interference then a small gap would be fine. Also to prevent capacitive changes between the coil and conductive ground, also a small gap is OK. However, to use the shield as a single turn TX there must be an overlap, otherwise the shield will act as a toroidal coil around the RX coil with little external field. Of course, we are relying on the conductive adhesive in the overlap to complete the circuit between turns, but it seems to be effective.

The shield does have two contacts with the circuitry. One end is ground and the other end to the drain of the switching mosfet as in the Spice circuits in earlier posts. I haven't used a series diode though, but I gather that it appears not to make any difference.

Eric.

I'm guessing the diode wouldn't make a difference with your circuit Tx=.5uH, is your coil in parallel with 2nf still has a calculated resonance of about 5MHz. I would be interested if adding a 5 or 10 ohm resistor across Tx for Rd effects the circuit when you do your tests. Thanks for starting this thread.

[KingJL]

I'm guessing the diode wouldn't make a difference with your circuit Tx=.5uH, is your coil in parallel with 2nf still has a calculated resonance of about 5MHz. I would be interested if adding a 5 or 10 ohm resistor across Tx for Rd effects the circuit when you do your tests. Thanks for starting this thread.

Green, Where did 2 nf come from? Given the circuit, discounting the effect of the fabric, I would expect the resonant freq to be in the range of 17-20 MHz assuming the coss of the IRFP220 @ ~150p and a gap capacitance of less than 5p (of course this capacitance would increase when he (if he does) adds cable). This is probably why the fabric is so effective at damping/absorbing/attenuating the flyback.