Ideal flyback voltage conditions

[mts]

I am designing and building a simple PI detector. I've scavenged the front end design from the multitude of great projects out here. I've tweaked it a bit to try and simplify it as much as possible. But I'm having a little trouble understanding what I should be shooting for in terms of optimal circuit characteristics. One such issue is the ideal flyback voltage. I'm using an IRF740 mosfet. The breakdown voltage of the mosfet is 400V (I've read out here that it is actually higher - see below).

I've read out here that the maxium target response will come from the maximum eddy currents being induced in the target. The maximum eddy currents occur when the current through the coil/mosfet is switched off as quickly as possible. This would also seem to imply that you would want the flyback voltage to be as high as possible (since the flyback voltage is a function of the derivative of the current through the coil). Isn't it true that the higher the flyback voltage the faster your mosfet is switching off? My flyback voltage appears to spike very quickly (within 1us) and goes to about 480V (which is probably the real breakdown voltage of the mosfet). I would think this would be ideal. However, I also read recently in this forum that you don't want the mosfet to enter breakdown as it hurts sensitivity/depth.

Is this true? Which is more important, high flyback voltage or making sure the mosfet doesn't enter breakdown? Or is there something else that I'm completely missing? Is the goal to aim for the highest possible flyback voltage without exceeding the breakdown voltage of the mosfet?

For what it's worth, I'm airtesting on a US quarter at about 7 inches which seems low to me (I'm using a 10" mono coil). There are many ways that I can likely increase this depth so I'm starting at the beginning (the transmit/receive circuit) and working my way forward. This will be a basic coin/relic machine.

Thanks for the help.

[Tinkerer]

I am designing and building a simple PI detector. I've scavenged the front end design from the multitude of great projects out here. I've tweaked it a bit to try and simplify it as much as possible. But I'm having a little trouble understanding what I should be shooting for in terms of optimal circuit characteristics. One such issue is the ideal flyback voltage. I'm using an IRF740 mosfet. The breakdown voltage of the mosfet is 400V (I've read out here that it is actually higher - see below).

I've read out here that the maxium target response will come from the maximum eddy currents being induced in the target. The maximum eddy currents occur when the current through the coil/mosfet is switched off as quickly as possible. This would also seem to imply that you would want the flyback voltage to be as high as possible (since the flyback voltage is a function of the derivative of the current through the coil). Isn't it true that the higher the flyback voltage the faster your mosfet is switching off? My flyback voltage appears to spike very quickly (within 1us) and goes to about 480V (which is probably the real breakdown voltage of the mosfet). I would think this would be ideal. However, I also read recently in this forum that you don't want the mosfet to enter breakdown as it hurts sensitivity/depth.

Is this true? Which is more important, high flyback voltage or making sure the mosfet doesn't enter breakdown? Or is there something else that I'm completely missing? Is the goal to aim for the highest possible flyback voltage without exceeding the breakdown voltage of the mosfet?

For what it's worth, I'm airtesting on a US quarter at about 7 inches which seems low to me (I'm using a 10" mono coil). There are many ways that I can likely increase this depth so I'm starting at the beginning (the transmit/receive circuit) and working my way forward. This will be a basic coin/relic machine.

Thanks for the help.

mts,

the Flyback voltage is not really important. What kicks the target, is the di/dt, that is the rate of change of the coil current or rather the magnetic field produced by the coil current.
Add a 0.1Ohm resistor in series with the coil winding, on the +V side, and look at the coil current.

Tinkerer

[mts]

What about limiting the flyback voltage so that the mosfet doesn't break down? Is that important or can I ignore that effect? In the HHII schematic Carl added a diode in series with the mosfet. My assumption was that this was designed to keep the mosfet from entering the breakdown region since the diode's breakdown voltage is lower than the mosfet (200V vs. 400V).

I understand that it is not the flyback voltage that does the dirty work in a PI system. But the flyback voltage is an indication of how quickly the current is turing off in the mosfet. It is also potentially an indication of how much current is going through the mosfet at the time of switch off. That's why my flyback voltage is probably higher than it should be. Even though I'm not switching off the mosfet as quickly as I would like to, the fact that I'm switching off such a large current (3-4A) makes up for this fact and sends my flyback voltage soaring.

I'm just trying to understand the tradeoffs and the reasoning behind putting in circuit elements to try and limit the flyback voltage for one reason or another. If it is a non-issue then I won't worry about it. But if having the flyback voltage exceed the breakdown voltage of the mostfet causes decreased depth or increased decay delays then I want to understand that.

Thanks!

[mikebg]

Mts,
Search in the forum a thread "newbie with PI questions". There are some answers.
The diagram below represents an experiment with different transistors. In intervals 2, 3 and 4 energy dissipates as heat in damping resistor, but in interval 3 energy dissipates also in the transistor. The instant power of dissipation in transistor is
P = I x Vbr, where "I" is coil current. If you connect a diode with Vbr=200V in series of a MOSFET with Vbr=400V, then you have total Vbr=600V and energy dissipates faster.

[porkluvr]

Why the diode.

I think the quick explanation is that the diode becomes reverse biased on the trailing edge of flyback and so acts as a small value capacitance in series between the coil and the FET capacitance.

It is advantageous to have as little capacitance in the circuit as possible for the fastest flyback.

Two capacitors in series will have a smaller net value than either capacitance alone (Ct=C1*C2/C1+C2) and the flyback will decay more quickly to bring the coil voltage into a range where sampling is feasible.

I don't understand why Carl would use a lower voltage diode but maybe he will clarify that and add more to the limited scope of my explanation. It seems to me that a 200PIV diode would be in jeopardy in a 400V flyback circuit.

I do not believe that increasing the breakdown voltage is the purpose of adding a series diode. It does not happen.

[Tinkerer]

mts, mikebg,

the matter of the Flyback voltage is only a part of the fundamental matter that you want to clarify. WHAT KICKS THE TARGET.
If you search the forum, for "what kicks the target" you will find many, many posts. I don't think a consensus has been reached yet. It is a complex matter.

The matter of the diode between the Mosfet and the coil is another such disputed matter. Ideally the diode should be of the same Vbr as the Mosfet. To be of any use, it needs to be an ultra fast switching diode.
If the Mosfet is chosen to have a Vbr high enough so that there is no avalanche breakdown, it will probably have a high drain capacitance. This capacitance extends the delay to the point where the Flyback decay has reached a level close to 0V, that is the traditional first sampling point.

An earlier sampling point means better sensitivity to smaller targets.

The series diode isolates this high drain capacitance and thus makes the detector more sensitive.
However, it is useless to use a 1N1007 or such kind of diode, that has a recovery time in the uS. The diode has to have a recovery time of a few nS to be of any use.
When using the diode, the Mosfet should not avalanche.
If you push the Flyback voltage to 1000V, you will have a faster decay, but you might also get insulation problems.

You talk of 3 to 4 Amps coil current. Would you give some information about the coil? Diameter, number of turns, dc resistance, uH, pulse length and voltage etc.?

Tinkerer

[mts]

Thanks for all of the help Tinkerer and Porkluvr. I've been reading as many articles on this forum as I can over the past couple of months. I guess that's given me just enough information to make me dangerous. However, these finer points still elude me. My degree is in EE but that was 20 years ago and I've been doing software development ever since. I've been going back and brushing up on discrete design but I've still got a long way to go. What you are both saying makes sense and I'll be sure to do some more reading. This place is a wealth of information and I could spend years here learning all of the finer details.

As for the coil current I'm using a 12v supply and the resistance of my coils is probably around 3 ohms. I can't measure it directly because I don't have a good enough multimeter at this time. But based on the mosfet having an on resistance of around 1 ohm (I'm currently only driving the gate with a 5V pulse so it is "on" but has a slightly higher resistance than the stated minimum) and using my oscilloscope, the voltage divider produced between the coil and mosfet confirms the current to be between 3 and 4 amps. So as I said in other posts, I'm getting 7" of depth on a quarter. But only because of brute force. I'd obviously like to reduce the current significantly to prolong battery life.

You can also calculate the approximate resistance from the length of wire I'm using for the coil. It's 20 turns of 24 guage stranded speaker wire. The coil is 10" in diameter. I've read through the excellent article on here about making fast coils and I know for a fact that my coil is far from fast. So there is some improvement that can be made in that regard.

But first things first. I want to work on improving the transmit circuit before moving on to other parts of the system.

Thanks again!

[bbsailor]

Thanks for all of the help Tinkerer and Porkluvr. I've been reading as many articles on this forum as I can over the past couple of months. I guess that's given me just enough information to make me dangerous. However, these finer points still elude me. My degree is in EE but that was 20 years ago and I've been doing software development ever since. I've been going back and brushing up on discrete design but I've still got a long way to go. What you are both saying makes sense and I'll be sure to do some more reading. This place is a wealth of information and I could spend years here learning all of the finer details.

As for the coil current I'm using a 12v supply and the resistance of my coils is probably around 3 ohms. I can't measure it directly because I don't have a good enough multimeter at this time. But based on the mosfet having an on resistance of around 1 ohm (I'm currently only driving the gate with a 5V pulse so it is "on" but has a slightly higher resistance than the stated minimum) and using my oscilloscope, the voltage divider produced between the coil and mosfet confirms the current to be between 3 and 4 amps. So as I said in other posts, I'm getting 7" of depth on a quarter. But only because of brute force. I'd obviously like to reduce the current significantly to prolong battery life.

You can also calculate the approximate resistance from the length of wire I'm using for the coil. It's 20 turns of 24 guage stranded speaker wire. The coil is 10" in diameter. I've read through the excellent article on here about making fast coils and I know for a fact that my coil is far from fast. So there is some improvement that can be made in that regard.

But first things first. I want to work on improving the transmit circuit before moving on to other parts of the system.

Thanks again!

mts,

The coil is part of the transmit system along with the value of the damping resistor.

With you being an EE, study coil current time constant rise times defined by coil inductance divided by total coil resistance (including coil resistance, MOSFET on resistance, and coax resistance). Generally, you want the coil TX on time to be near 3 time constants (TC).

The TX off time constant is defined by the inductance divided by the value of the damping resistor. Generally, you want the turn off slope to be as close to vertical as possible. The closer to vertical, the faster the turn off and the higher the flyback voltage.

Generally, to fully stimulate small targets (like small gold nuggets) you want the coil turn off TC to be 5 times faster than the target TC. So, fast coils become more important for smaller targets.

When you study these variables and play with this for a while you will see how all these things relate to each other.

I'm glad you enjoyed the fast coil article.

bbsailor

[mts]

Thanks bbsailor. That all makes sense. I'm going to have to go back to the drawing board on my coil if I want to be able to detect gold nuggets. The original expectation was that this was going to be a coin/relic machine. But the more I get into this thing the more I find that I'm wanting to go for the gold!

[bklein]

mts,

the Flyback voltage is not really important. What kicks the target, is the di/dt, that is the rate of change of the coil current or rather the magnetic field produced by the coil current.
Add a 0.1Ohm resistor in series with the coil winding, on the +V side, and look at the coil current.

Tinkerer

Is there evidence of the target being "kicked" by the first energizing of the TX coil? Say you didn't care about power drain and you just energized the TX coil and had a separate RX coil to analyze the target and could somehow prevent TX coil induction into the RX coil (or null it).

Barry

[Tinkerer]

Is there evidence of the target being "kicked" by the first energizing of the TX coil? Say you didn't care about power drain and you just energized the TX coil and had a separate RX coil to analyze the target and could somehow prevent TX coil induction into the RX coil (or null it).

Barry

Hi Barry,

Look at the TINKERERS_V1 discriminating IB-PI, it does just as you say.

The first energizing of the TX coil produces a very good response that is of one polarity for iron and the opposite polarity for gold.

http://www.geotech1.com/forums/showp...24&postcount=6

Tinkerer

[bklein]

Hi Barry,

Look at the TINKERERS_V1 discriminating IB-PI, it does just as you say.

The first energizing of the TX coil produces a very good response that is of one polarity for iron and the opposite polarity for gold.

http://www.geotech1.com/forums/showp...24&postcount=6

Tinkerer

This seems to be an Eagle schematic, right?
I am uncommitted on a layout program....
Has anyone tried a phase reversal of the TX coil then?
Barry

[taliesin]

why did everyone stop here ?

this was good reading .and everyone stopped .

[DOOLEY]

it WAS ,and then it stopped , i was just getting into that , scrolling down and hoping the holy grail was at the bottom , the only thing there was the end.

i'm currently experimenting with irf840's , found on ebay 10 for £8 bargain.

8A cont
32A pullsed
500V (600V under test)
0.75 ohm on res.

slightly higher resistance than an irf740 but 0.25 ohm extra realy worth worrying about ?

anyway back to flyback , i'm getting 600V spike with the irf840 , at which point , i suspect internal zener is limiting it.

was using a humble irf510 before with 100v (nearer 150V) limit ,

so little experiment , using same coil , setup etc...etc... tried the irf540 and the irf840 in same circuit. and using a flan tin bottom (steel) on the coil , central .

i know this sounds simple / crazy and amaturish , but noticed quite early on that metal objects made a noise / vibrated the nearer the coil they got

one thing i noticed between the 2 mosfets was that physical coil / tin noise is allot louder with the irf840 , indicating that high flyback is very important , suspect this is what is "kicking the target"

if so then the higher the flyback then better.

this all makes sense , as if you "kill" the flyback using a diode over the coil the noise output drops , allot ,very quiet ,

so my conclusion , more flyback , more vibrating / noise of target , in this case a flan tin bottom , =more eddy currents in target = more reaction to voltage decay curve on coil output.

am i wrong ??

[AdrianM]

Well, it is a very interesting but short thread so why don't we revive it! I would like to consider some variations on damped flyback as well - Tinkerer has shown a number of these, but the one that interests me most is that the simple addition of a capacitor in parallel with the coil modifies the flyback into a half-sine wave for example.

It may or may not be obvious that this wouldn't be useful in a mono coil PI. After all we normally try to rid the coil of parasitic C. However, It certainly has application in TEM and induction balance designs. The rate of change of coil current falls as a result of the work done in charging the capacitor, but this may be the best way to stimulate longer time constant targets. It also has the great advantage of making the flyback peak voltage controllable such that mosfet avalanche doesn't take place. I'm in the process of finding out exactly what difference it makes to target reaction.