PI Monocoil - sampling under 4us almost impossible?

[B^C]

I agree with Simon that were all on different pages, especially me.

The rate of current fall in an inductor is V/L, as i read things & maybe i am reading things wrong but it seems as though people are trying to read target response after the flyback returns to 0?

Put all capacitance, voltage etc aside for one moment, it is the decay of the spike we need to read.

If our spike settles at say 1uS no target--unstable detector area, then when metal etc is introduced to the coil the spike decay is longer & this is what we read.

Maybe that's what is being talked about, but at times it seems different with talk of TC of targets etc?

I was drawing a diagram to explain & then i found one by drawn by Reg, i hope he doesn't mind me posting it here? Anyway thanks Reg!

Like i said i may be reading things wrong so i thought i would add this to make sure were talking about the same things.

[Tinkerer]

Tinkerer,

I address your very good questions that I put in bold above. I will put this in terms that are observable with a little simple math.

Look at the Hammerhead mono coil circuit for my example below. Just substitute your own values and it should work out the same as my example. I will try to write so you can visualize what is going on.

1. MOSFET over voltage will cause the flyback voltage to clamp at the MOSFET voltage and will tend to extend the coil discharge time by the horizontal length of the flyback flat top. If this is less than 1us it will have a minimal effect but this does tend to cause the MOSFET to heat up if the flat-top clamping is too long. With the wide voltage ratings of MOSFETs this should not be a problem. One way to reduce the voltage across the MOSFET is to put a resistor in series with the MOSFET that acts as a voltage divider. Generally, a 1 to 2 ohm resistor is adequate for a high power PI design.

2. This part gets a little tricky so bear with me and try to draw this out on paper to help visualize what is going on. In a mono coil circuit design, the value of Rd is actually in parallel with R12 (the Hammerhead opamp input resistor) when the clamping diodes are conducting above 0.6 to 0.7V depending on the actual diode bias voltage. I will use a voltage of 0.65 to represent an average diode voltage for this example. If Rd is 680 ohms (as in the Hammerhead) the discharge curve occurs in two stages:
Stage 1 is the discharge curve above 0.65V with Rd and Rin in parallel or 404.76 ohms. This is calculated by: Rd X Rin divided by Rd + Rin or the classic parallel resistor formula. The slope of the discharge is governed by the coil inductance (500uh for the Hammerhead) divided by 404.76 ohms or 1.235us down to 0.65v.
Stage 2 occurs below 0.65V is is only governed by the value of Rd alone as the clamping diodes ar no longer placing Rin in parallel with Rd. Stage 2 begins a new steeper slope of 500 divided by 680 or 0.735us down to the point where the opamp comes out of saturation. A lower opamp gain will help to speed this up as well as increasing the supply voltage to the opamp.

The earliest sampling occurs at the optimal damping point where the opamp comes out of saturation and a target signal can be sampled in the RX window. A "lock-up" will occur if you try to push the sampling too early, before the earliest no-target flat spot occurs. Otherwise, you will be sampling the tail end of the coil oscillations before they are fully damped and loose most sensitivity.

Now for the rest of the story.

The combined capacitance of the TX circuit governs that adjusted value of Rd. Put a 7 ft long coax on the coil and you may have almost 200pf of capacitance to cope with. Reduce the cable length to 2.5 ft and use low capacitance cable and you may only have about 50pf to deal with. This capaitance reduction should allow you to adjust Rd to about a 200 ohm higher value. However this new Rd value with the shorter cable is now 880 ohms but it is still in parallel with Rin so it looks like a parallel value 468 ohms with a discharge TC of 500/468 or 1.068us for stage 1 and 500/880 or 0.568us for stage 2 of the coil discharge curve. This should improve the minimum delay by about 2us above 10 us delay.

Why is Scotch24 a good shield? It is a thin wire mesh that has less surface area than a solid shield material. It functions like a leaky coax or capacitor. They use leaky coax cable to allow radio signals and cell phones to operate in tunnels.

If you wind a coil with thicker stranded wire such as AWG 22 to 26, the wire bundle will be thicker than a coil wound with AWG 30 wire. The thinner wire bundle of AWG 30 wire will have less capacitance than a thicker wire bundle. The actual measured coil-to-shield capacitance is not fully imposed on increasing the total RX circuit capacitance. Only about 20% of the coil-to-shield capacitance is imposed on the total TX circuit capacitance. Remember, it is the sum total of a few factors that governs capacitance and ultimately the value of Rd.

Now for the more subtle factors. Bear with me and draw this out. Rd and Rin form a voltage divider in stage 2 when the RX sample occurs. Look at the combined value of Rin (1K) and Rd (680) as a voltage divider pot with a total resistance of 1680 ohms. The target sample signal level is split by the ratio of Rd to R total or 1680 divided by 680 or about 2.5 attenuation.

If you use a MOSFET drain diode and isolate the TX circuit from the MOSFET COSS than you can set Rd to a higher value. Let's say that we can get Rd to be 1K. This reduces the attenuation to a ratio of 2 (Rd of 1000 + Rin of 1000 or 2000 divided by 1000 or 2). This make a more sensitive coil with less signal attenuation.

This all gets down to one simple question. What is the TC of the smallest target I want to detect with my mono coil? The coil discharge TC should be 5 times faster than the TC of the smallest target you want to detect to fully stimulate it. This is why when you get below 10us attending to these capacitance lowering factors can yield an observable improvement but only for those small, low TC targets.

However, not all locations require a faster coil. If you are a beach or coin hunter, you do not want to go too much below 10us as wet sand becomes one large target. What if you make a 300uH coil that can sample at 7us? You have 3us of what I call "reserve sensitivity". Add 2 to 4 turns to the coil and now sample at near 10us. This makes the coil a little more sensitive to coins and a small bit slower but still within the 10us range for beach use.

So far the best way to make a low capacitance coil is to do a few things that all collectivly work in the same direction to reduce total TX circuit capacitance.
1. Use PTFE (Teflon) insulation on your wire
2. Minimize coil-to-shield capacitance with a spacer and by using Scotch24
3. Use the lowest capacitance and shortest coax (17pf/ft is the lowest I have seen)
4. Keep the pulse width low enough to not exceed the MOSFET voltage very much or add a series resistor (a few ohms) to minimize the voltage across the MOSFET
5. Minimize MOSFET COSS by using a high speed drain diode with a discharge resistor or use a low COSS MOSFET.

Ideally, the coil discharge time constant should be verticle but this is impossible with the physics that we all know. Getting the coil discharge as verticle as possible is simply a matter of managing coil and circuit capacitance with less capacitance making a more vertical coil discharge TC. The net result of everything that you do to minimize capacitance is summarized in the final adjusted optimum value of Rd which then ripples out to other areas of analysis as I have pointed out above.

I hope I have made this subject a little more visually understandable and helped you to form a mental model to anticipate what is happening when you make coil or circuit alterations.

bbsailor

Thanks bbsailor,

your analytical mind is always the best. We tend to look at a single factor, when there are many. Thanks for bringing us back to the reality.

Tinkerer

[Tinkerer]

Ok -- I basically meant you don't have to work so hard to shorten the flyback pulse in order to get a clear space for sampling -- your IB design nulls the flyback pulse so it can't interfere.


I still feel it is all relative - a lower energy pulse does not decay faster relative to it's energy - and the target response is proportional to it's energy. So pulse is smaller, target is smaller, you have to wait just as long for target to emerge from pulse. Does that seem reasonable?

Cheers,

-SB

Right about the first part. It is easier with the IB coil.

For the lower energy pulse, hmmm, looking at the formula, there is no provision for the energy. One of the missing links.
the easy way is to open MiscEl, - Inductor discharge - enter 1A, 300u, 500R
and look at the discharge curve at 4uS. the coil voltage is at 620mV.
Repeat with 4A, 300u, 500R and the coil voltage is at 2.4V.

Now you also need to look at the Flyback voltage, at 4A coil current it goes to 2kV. You see that we might have a small problem there.
Therefore we also will have to change the R value of the damping resistor.
Now you really see the change in time it takes to discharge the same coil with more energy stored.
This does not take into account the capacitance. Could somebody please come up with that formula?
When you use the Capacitor discharge window on MiscEl, you see that the capacitor discharge time is not the same as the inductor discharge. This is what causes the ringing, when the damping resistor is too high.
I love MiscEl,

Tinkerer

[simonbaker]

Simonbaker,

Quote:
Tinkerer seems to be pursuing the "IB" dual coil variation of PI detector, where I think shorter flyback pulse is irrelevant to sampling since you can sample whenever you want. End Quote

Not quite right. The IB coil configuration allows you to sample during the TX pulse, but for short TC targets you still have to sample as near the transient as possible.

Ok -- I basically meant you don't have to work so hard to shorten the flyback pulse in order to get a clear space for sampling -- your IB design nulls the flyback pulse so it can't interfere.

Nicolae's question leaves out an important factor. The ground response.
Ignoring the ground response for the moment, a lower energy pulse decays faster.

I still feel it is all relative - a lower energy pulse does not decay faster relative to it's energy - and the target response is proportional to it's energy. So pulse is smaller, target is smaller, you have to wait just as long for target to emerge from pulse. Does that seem reasonable?

Cheers,

-SB

[Tinkerer]

I'm a beginner too, so I really appreciate the basic questions.

I think your ideas make a lot of sense -- around here a lot of differences in opinion are simply because it's too tedious to really define our conditions exactly so we all are on different pages half the time! And we have different backgrounds -- makes for lively discussions.

I agree it seems that once you have a coil made, there is only one optimal Rd (unless you play tricks with underdamping) -- I think your point is, what's left to play with? -- answer: pulse width and sampling delay.

So maybe you are leading up to the idea of a metal detector that explicitly uses controllable pulse width and sample delay to eke out more information about what's underground. I would bet people somewhere look at that extensively, but still sounds like a good area for experiments and design.

Tinkerer seems to be pursuing the "IB" dual coil variation of PI detector, where I think shorter flyback pulse is irrelevant to sampling since you can sample whenever you want.

But to sum up the various messages responding to your thoughts, it seems most people say you have to reduce the capacitance in order to shorten flyback pulse by increasing Rd, so no disagreement with your statement that Rd is not adjustable by itself.

Now, on your idea of shortening TX pulse to sample sooner -- it seems to me that won't buy you anything. The flyback pulse looks shorter, but it really isn't, in a relative sense -- it is just smaller in amplitude, but has the same decay shape. The target response is also smaller, so you will have to wait the same time (as with the larger pulse) until the smaller target emerges from the smaller flyback pulse. I didn't do any math on that, but that's my gut feeling.

Keep throwing those questions out, that's how those of us catching up get a chance to learn.

Cheers,

-SB

Simonbaker,

Quote:
Tinkerer seems to be pursuing the "IB" dual coil variation of PI detector, where I think shorter flyback pulse is irrelevant to sampling since you can sample whenever you want. End Quote

Not quite right. The IB coil configuration allows you to sample during the TX pulse, but for short TC targets you still have to sample as near the transient as possible.

Nicolae's question leaves out an important factor. The ground response.
Ignoring the ground response for the moment, a lower energy pulse decays faster.
You can produce a lower energy pulse by adding DC resistance to the coil, or by shortening the TX pulse.

This is for a mono coil. With the IB coil things are different.

Nicolae,

have you considered trying a "mono" coil that has a separate RX winding? This is a way to reduce the delay.

Tinkerer

[simonbaker]

As you know, I am very much a beginner in the metal detector world. And at some stage I realised something which may be counterintuitive to some of us. Let't say we have a metal detector and the only changes we can do to it is to change the main pulse width (which changes the flyback pulse voltage) and we can also change the delay of the first sample.
Let's assume we can change the pulse width of the metal detector between 50 and 100us.
The fact that for this adjustment the transistor reaches breakdown voltage or not is not important.
Let's say you have a target (aluminium foil) very close to the coil (or very close to the surface of the ground - it could be a few cm under the ground if you wish). You set the main pulse width to 75us and optimize the first sample point. And you start reducing the size of the target to the extent where you are hardly able to detect that sample.

Now, reduce the size of the sample even more, so you can't detect it anymore.
We set the initial conditions for the experiment.

Now, what do you have to do in order to detect the sample again? Will you be able to detect the sample by changing the pulse width toward 100us or toward 50us and reoptimize the first sample delay? Increasing the pulse width toward 100us will increase the flyback voltage, but also the decay time.

I would decrease the pulse width and readjust the first sample delay (the flyback voltage will be lower!) in order to detect that smaller target. But let us hear what other people would do and if anybody thinks I am crazy

Regards,
Nicolae

I'm a beginner too, so I really appreciate the basic questions.

I think your ideas make a lot of sense -- around here a lot of differences in opinion are simply because it's too tedious to really define our conditions exactly so we all are on different pages half the time! And we have different backgrounds -- makes for lively discussions.

I agree it seems that once you have a coil made, there is only one optimal Rd (unless you play tricks with underdamping) -- I think your point is, what's left to play with? -- answer: pulse width and sampling delay.

So maybe you are leading up to the idea of a metal detector that explicitly uses controllable pulse width and sample delay to eke out more information about what's underground. I would bet people somewhere look at that extensively, but still sounds like a good area for experiments and design.

Tinkerer seems to be pursuing the "IB" dual coil variation of PI detector, where I think shorter flyback pulse is irrelevant to sampling since you can sample whenever you want.

But to sum up the various messages responding to your thoughts, it seems most people say you have to reduce the capacitance in order to shorten flyback pulse by increasing Rd, so no disagreement with your statement that Rd is not adjustable by itself.

Now, on your idea of shortening TX pulse to sample sooner -- it seems to me that won't buy you anything. The flyback pulse looks shorter, but it really isn't, in a relative sense -- it is just smaller in amplitude, but has the same decay shape. The target response is also smaller, so you will have to wait the same time (as with the larger pulse) until the smaller target emerges from the smaller flyback pulse. I didn't do any math on that, but that's my gut feeling.

Keep throwing those questions out, that's how those of us catching up get a chance to learn.

Cheers,

-SB

[bbsailor]

hobbes,

this is probably the crisp and cleanest explanation I have seen. You make an excellent professor.

How can a Tinkerer add to that? Tinkering is based on practical observation. (All I have left if I am too stupid to do Math.)
Now lets see a few spots where we can reduce capacitance. It is not as easy as seems, because for some reason I don't understand, some of the theoretical capacitance does not show up.
Mosfet Coss.
The right kind of diode isolates that. It is funny to see all the simulations and simulators fighting over that one. Just add the diode in a real circuit. Push the Flyback to close to the Mosfet break down, but not over it. Then look at the Voltage on the drain. There you see that the Flyback Voltage is stored by the drain capacitance.
Cable capacitance:
It is easy to get 100pf cable capacitance. Now consider charging this 100pf capacitor to 500V and then discharging it. It takes time.
If you put the damping resistor at the coil, the cable will not see that much of the Flyback. How about putting the 1K input resistor and the limiting diodes right next to it?
Coil shielding:
Yes it adds a considerable amount of capacitance. How can we tackle that?
What if we made this shield capacitor a very leaky one. It would discharge fast no?
Graphite paint I believe there is a trade mark Aquadag? can be applied to have a certain resistance. A leaky capacitor?
By the way, I have posted recipes of how to make graphite paint several times. the secret ingredient is the fumed Silica. It regulates the conductivity of the shield.

Tinkerer

Tinkerer,

I address your very good questions that I put in bold above. I will put this in terms that are observable with a little simple math.

Look at the Hammerhead mono coil circuit for my example below. Just substitute your own values and it should work out the same as my example. I will try to write so you can visualize what is going on.

1. MOSFET over voltage will cause the flyback voltage to clamp at the MOSFET voltage and will tend to extend the coil discharge time by the horizontal length of the flyback flat top. If this is less than 1us it will have a minimal effect but this does tend to cause the MOSFET to heat up if the flat-top clamping is too long. With the wide voltage ratings of MOSFETs this should not be a problem. One way to reduce the voltage across the MOSFET is to put a resistor in series with the MOSFET that acts as a voltage divider. Generally, a 1 to 2 ohm resistor is adequate for a high power PI design.

2. This part gets a little tricky so bear with me and try to draw this out on paper to help visualize what is going on. In a mono coil circuit design, the value of Rd is actually in parallel with R12 (the Hammerhead opamp input resistor) when the clamping diodes are conducting above 0.6 to 0.7V depending on the actual diode bias voltage. I will use a voltage of 0.65 to represent an average diode voltage for this example. If Rd is 680 ohms (as in the Hammerhead) the discharge curve occurs in two stages:
Stage 1 is the discharge curve above 0.65V with Rd and Rin in parallel or 404.76 ohms. This is calculated by: Rd X Rin divided by Rd + Rin or the classic parallel resistor formula. The slope of the discharge is governed by the coil inductance (500uh for the Hammerhead) divided by 404.76 ohms or 1.235us down to 0.65v.
Stage 2 occurs below 0.65V is is only governed by the value of Rd alone as the clamping diodes ar no longer placing Rin in parallel with Rd. Stage 2 begins a new steeper slope of 500 divided by 680 or 0.735us down to the point where the opamp comes out of saturation. A lower opamp gain will help to speed this up as well as increasing the supply voltage to the opamp.

The earliest sampling occurs at the optimal damping point where the opamp comes out of saturation and a target signal can be sampled in the RX window. A "lock-up" will occur if you try to push the sampling too early, before the earliest no-target flat spot occurs. Otherwise, you will be sampling the tail end of the coil oscillations before they are fully damped and loose most sensitivity.

Now for the rest of the story.

The combined capacitance of the TX circuit governs that adjusted value of Rd. Put a 7 ft long coax on the coil and you may have almost 200pf of capacitance to cope with. Reduce the cable length to 2.5 ft and use low capacitance cable and you may only have about 50pf to deal with. This capaitance reduction should allow you to adjust Rd to about a 200 ohm higher value. However this new Rd value with the shorter cable is now 880 ohms but it is still in parallel with Rin so it looks like a parallel value 468 ohms with a discharge TC of 500/468 or 1.068us for stage 1 and 500/880 or 0.568us for stage 2 of the coil discharge curve. This should improve the minimum delay by about 2us above 10 us delay.

Why is Scotch24 a good shield? It is a thin wire mesh that has less surface area than a solid shield material. It functions like a leaky coax or capacitor. They use leaky coax cable to allow radio signals and cell phones to operate in tunnels.

If you wind a coil with thicker stranded wire such as AWG 22 to 26, the wire bundle will be thicker than a coil wound with AWG 30 wire. The thinner wire bundle of AWG 30 wire will have less capacitance than a thicker wire bundle. The actual measured coil-to-shield capacitance is not fully imposed on increasing the total RX circuit capacitance. Only about 20% of the coil-to-shield capacitance is imposed on the total TX circuit capacitance. Remember, it is the sum total of a few factors that governs capacitance and ultimately the value of Rd.

Now for the more subtle factors. Bear with me and draw this out. Rd and Rin form a voltage divider in stage 2 when the RX sample occurs. Look at the combined value of Rin (1K) and Rd (680) as a voltage divider pot with a total resistance of 1680 ohms. The target sample signal level is split by the ratio of Rd to R total or 1680 divided by 680 or about 2.5 attenuation.

If you use a MOSFET drain diode and isolate the TX circuit from the MOSFET COSS than you can set Rd to a higher value. Let's say that we can get Rd to be 1K. This reduces the attenuation to a ratio of 2 (Rd of 1000 + Rin of 1000 or 2000 divided by 1000 or 2). This make a more sensitive coil with less signal attenuation.

This all gets down to one simple question. What is the TC of the smallest target I want to detect with my mono coil? The coil discharge TC should be 5 times faster than the TC of the smallest target you want to detect to fully stimulate it. This is why when you get below 10us attending to these capacitance lowering factors can yield an observable improvement but only for those small, low TC targets.

However, not all locations require a faster coil. If you are a beach or coin hunter, you do not want to go too much below 10us as wet sand becomes one large target. What if you make a 300uH coil that can sample at 7us? You have 3us of what I call "reserve sensitivity". Add 2 to 4 turns to the coil and now sample at near 10us. This makes the coil a little more sensitive to coins and a small bit slower but still within the 10us range for beach use.

So far the best way to make a low capacitance coil is to do a few things that all collectivly work in the same direction to reduce total TX circuit capacitance.
1. Use PTFE (Teflon) insulation on your wire
2. Minimize coil-to-shield capacitance with a spacer and by using Scotch24
3. Use the lowest capacitance and shortest coax (17pf/ft is the lowest I have seen)
4. Keep the pulse width low enough to not exceed the MOSFET voltage very much or add a series resistor (a few ohms) to minimize the voltage across the MOSFET
5. Minimize MOSFET COSS by using a high speed drain diode with a discharge resistor or use a low COSS MOSFET.

Ideally, the coil discharge time constant should be verticle but this is impossible with the physics that we all know. Getting the coil discharge as verticle as possible is simply a matter of managing coil and circuit capacitance with less capacitance making a more vertical coil discharge TC. The net result of everything that you do to minimize capacitance is summarized in the final adjusted optimum value of Rd which then ripples out to other areas of analysis as I have pointed out above.

I hope I have made this subject a little more visually understandable and helped you to form a mental model to anticipate what is happening when you make coil or circuit alterations.

bbsailor

[B^C]

Oh yeah i forgot, that damping resistor that everybody keeps talking about is an old way of achieving an effect, throw it in the rubbish & have a look at a different way. There is more than one way to skin a cat!

[B^C]

Gday Nicolae,

I don't think having a differences is pissing people off, well not me anyway.

Healthy discussion is what is needed.

I think your right in one sense in that the same old hum drum circuits & the old way of thinking just does not do the job, this is why it needs a new way of thinking & circuits to cope with what is "actually needed".

I am looking at a circuit now & the flyback on the scope is out of sight & it has a great inductive kick & it's the best detector i have used in "real world" conditions.
Put the old ways in the closet & have a look at what's needed.

If physics or science can explain exactly what we need in a detector then how come we haven't got one worth squat?
If it was that easy then everybody would have already done it.

In ideal situations physics or science may be able to tell what we need but now lets have a look at it where we need to use it & things change dramatically, throw the text books out & have a new look at the problem.

Do any of us know exactly what we need---nope, but by following the good ol boys is only going to give us what they had & obviously it wasn't & still isn't good enough.

[nick_f]

Detecting smaller samples with any metal detector

As you know, I am very much a beginner in the metal detector world. And at some stage I realised something which may be counterintuitive to some of us. Let't say we have a metal detector and the only changes we can do to it is to change the main pulse width (which changes the flyback pulse voltage) and we can also change the delay of the first sample.
Let's assume we can change the pulse width of the metal detector between 50 and 100us.
The fact that for this adjustment the transistor reaches breakdown voltage or not is not important.
Let's say you have a target (aluminium foil) very close to the coil (or very close to the surface of the ground - it could be a few cm under the ground if you wish). You set the main pulse width to 75us and optimize the first sample point. And you start reducing the size of the target to the extent where you are hardly able to detect that sample.

Now, reduce the size of the sample even more, so you can't detect it anymore.
We set the initial conditions for the experiment.

Now, what do you have to do in order to detect the sample again? Will you be able to detect the sample by changing the pulse width toward 100us or toward 50us and reoptimize the first sample delay? Increasing the pulse width toward 100us will increase the flyback voltage, but also the decay time.

I would decrease the pulse width and readjust the first sample delay (the flyback voltage will be lower!) in order to detect that smaller target. But let us hear what other people would do and if anybody thinks I am crazy

Regards,
Nicolae

[nick_f]

[quote=B^C;91028]Gday Guy's,

Nicolae might not be seeing much of a change due to the amperage changing at the same rate each test?.
Or there may be to much capacitance, or the flyback is not being "forced" enough to see an actual difference, remember it would take an infinite voltage for the current to drop in an instant.

Now i think Nicolae may have been seeing the current drop quicker with little flyback because this can be the case if the flyback is not high enough.


It seems to me that the more capacitance you have the higher the flyback has to be to compensate.[/quote]


Hi B^C,

I am afraid I will have to disagree with you in a few areas. I consider all the above statements I marked with bold characters to be incorrect or incomplete. All these statements are actually based on your belief that a higher flyback leads to a faster decay. As Aziz pointed out, it is very important to state if you keep the energy constant in the system and compare actually two different systems (different capacitances or inductances). Only if you change the inductance or capacitance you can get a higher flyback voltage, for the same energy supplied to the system. In that case, Azis is right and you are right too.

But, as in our practical case we have a metal detector already built and we don't change L and C in circuit. Even more, we take care not to reach the breakdown voltage on the transistor.
We can only change the level of flyback pulse by changing the maximum current before we turn off the transistor (which we do by changing the pulse width). This is the situation I was always talking about, and in this case, higher flyback voltage leads to an increase of the decay time. This certainly means the energy supplied to the system increases as well.

Aziz and Hobbes presented the formula for the critically damped circuit. In order to get the shortest decay time of the voltage, there is only one value for the damping resistor and that has nothing to do with the flyback voltage. In the case of the underdamped circuit, we get the voltage of the flyback pulse reach zero sooner than in the case of the critically damped circuit. The problem is the voltage won't remain on zero, it will keep going passed zero up and down and eventually it will settle to zero, but after a longer amount of time than for the critically damped circuit.

The flyback voltage can be 1V or 1million volt and the shortest decay will be obtained still for the same value of the damping resistor.
If this is the case, then bbsailor may not be right (he said there is a limited range for pulse width that maintains the same damping resistance - as long as we don't reach breakdown voltage of the mosfet, that should not be the case). He might be right if the transistor in the system has a variable capacitance with the voltage (like a varicap diode) - we need to do some practical testing to see if that is the case.

I found a link that shows some graphs for the critically damped circuit, an overdamped circuit and underdamped circuit: http://www.coilgun.info/theoryinduct...oscillator.htm

Regards,
Nicolae

P.S. - I am sorry for pissing everybody off with my stuborness and lack of understanding

[Tinkerer]

Using ideal components, the coil and damping resistor can be analyzed as a parallel RLC circuit. Critical damping occurs when Q = 1/2. If we take the equation for Q in a parallel RLC circuit:



insert a value of 1/2 for Q and then solve for R, here's what we get:



This formula demonstrates the relationship of C to R. An increase in the value of C requires a reduction in the resistance of R for critical damping.

Here are a few examples:

If we solve for L=300uH and C=50pF, then R=1224.74 ohms for critical damping.

If we solve for L=300uH and C=100pF, then R=866.03 ohms for critical damping.

If we solve for L=300uH and C=250pF, then R=547.72 ohms for critical damping.

hobbes,

this is probably the crisp and cleanest explanation I have seen. You make an excellent professor.

How can a Tinkerer add to that? Tinkering is based on practical observation. (All I have left if I am too stupid to do Math.)
Now lets see a few spots where we can reduce capacitance. It is not as easy as seems, because for some reason I don't understand, some of the theoretical capacitance does not show up.
Mosfet Coss.
The right kind of diode isolates that. It is funny to see all the simulations and simulators fighting over that one. Just add the diode in a real circuit. Push the Flyback to close to the Mosfet break down, but not over it. Then look at the Voltage on the drain. There you see that the Flyback Voltage is stored by the drain capacitance.
Cable capacitance:
It is easy to get 100pf cable capacitance. Now consider charging this 100pf capacitor to 500V and then discharging it. It takes time.
If you put the damping resistor at the coil, the cable will not see that much of the Flyback. How about putting the 1K input resistor and the limiting diodes right next to it?
Coil shielding:
Yes it adds a considerable amount of capacitance. How can we tackle that?
What if we made this shield capacitor a very leaky one. It would discharge fast no?
Graphite paint I believe there is a trade mark Aquadag? can be applied to have a certain resistance. A leaky capacitor?
By the way, I have posted recipes of how to make graphite paint several times. the secret ingredient is the fumed Silica. It regulates the conductivity of the shield.

Tinkerer

[hobbes_lives]

Using ideal components, the coil and damping resistor can be analyzed as a parallel RLC circuit. Critical damping occurs when Q = 1/2. If we take the equation for Q in a parallel RLC circuit:



insert a value of 1/2 for Q and then solve for R, here's what we get:



This formula demonstrates the relationship of C to R. An increase in the value of C requires a reduction in the resistance of R for critical damping.

Here are a few examples:

If we solve for L=300uH and C=50pF, then R=1224.74 ohms for critical damping.

If we solve for L=300uH and C=100pF, then R=866.03 ohms for critical damping.

If we solve for L=300uH and C=250pF, then R=547.72 ohms for critical damping.

[simonbaker]

Hi Aziz,
We should never reach the breakdown voltage of the transistor. Once we reach that, it makes no sense to pump more current through the coil, because it will only be wasted battery without any benefit to the detector.
If the breakdown voltage is reached for 1us, that 1us will only add to the decay time, without any advantage.
If we assume an equal amount of energy and the same inductor, and same parasitic capacitance, how can we generate a higher flyback voltage? We can't increase damping resistor, because the circuit will get underdamped. I would say that equal energy and higher flyback voltage for the same circuit is not possible (definition of madness: keep doing the same thing and expect a different result ) - sorry, I was just joking here).
In theory, we could obtain a higher flyback voltage for the same amount of energy by increasing the quality factor of the coil (decreasing the series resistance in circuit).

Regards,
Nicolae

I think I agree -- flyback voltage is proportional to how fast we can stop the current. The limiting factor must be the capacitor, because in ideal case with no capacitance, we could increase resistor and make voltage spike go towards infinity.

So I think you just have to get capacitance down, otherwise fooling with Rd to no avail.

Aziz seems to be talking about different case, where MOSFET breakdown is limiting the flyback voltage, so naturally a higher breakdown would help.

I think I agree also that going into breakdown is non-productive because di/dt seems to be zero -- wouldn't that make a double "pulse" as seen by the target, one pulse before MOSFET breakdown and one after? But maybe two quick pulses give a useful kick to typical targets for detection purposes.

I have another question about differential design, which I'll ask in another message...

Cheers,

-SB

[B^C]

Gday Guy's,

I have been following this with interest, the inductor calculation
V= L dI/dt may solve the issue. Someone with a bit more mathematical skill may be able to write out a few examples.
Belt me over the head at any time if it makes no sense!
A capacitor discharges current & the voltage ramps to 0 & an inductor discharges voltage & the current ramps to 0. So we need to discharge voltage for the inductor.
An inductor holds it current at 0 terminal voltage & the capacitor holds it voltage at 0 current, so the higher the voltage spike of the inductor the quicker the current will fall.

The flyback is determined by the duty cycle & the amperage not so much total amps but how fast the amperage changes direction, take note of this fact.

Nicolae might not be seeing much of a change due to the amperage changing at the same rate each test?.
Or there may be to much capacitance, or the flyback is not being "forced" enough to see an actual difference, remember it would take an infinite voltage for the current to drop in an instant.

Force the flyback hard & make it discharge when you want it to not when it wants to & you will see a difference & don't cut it short, let the sucker fly to drop the current.

Now i think Nicolae may have been seeing the current drop quicker with little flyback because this can be the case if the flyback is not high enough.

But!
We require a good "inductive kick" & if we just let the flyback just go "zing---fizzz----plop" like a chinese missile attempt well we can expect no difference or if the flyback just rolls off the edge & falls over we are not going to get a good inductive kick.
Kick it in the backside & let it fly, it's called flyback not zing-plop, if you can relate to what i'm saying without the inductive kick we may detect a car at 12".

A good storage of current is required to kick this in the butt when we "need" it.

If all the neccessary ingredients we need, like the duty cycle, the amperage, the amperage change of direction & then the final "kick" are tuned right you will see good flyback & the current drop fast as well.
If all things remain the same & she's a fizzer then you just will not see it!

I have attached a screen shot of forced flyback for want of a better word.

The Light Green--Lime is the switch off, the Red line is the Inductor Amperage, Dark Blue is the Inductor Voltage & the Light Blue is what i get after the parrallel diodes with the circuit i have at the moment.

It seems to me that the more capacitance you have the higher the flyback has to be to compensate.