Looking for a solution to improve slow coil electronically to let it decay fast and keep R-dump high

[greylourie]

I'm sure I heard some talking of slight under damping sometimes being an advantage. Was it you Waikikii ? Anyway, there might be some tips to be found in HDPhilips two threads on his uno32 and his experiments with the QED hardware. He might have some scope shots posted showing how the damping behaves.

[Waikiki_Sweep]

It was really interesting reading. Thank you.
I only would like to fix some moments. According my believes:

Join DateFeb 2005LocationBarnegat, NJPosts338

The larger the shunt resistor, the greater the current dissipation and the faster the current decay.

The larger the shunt resistor, the smaller the current dissipation and the slower the current decay.

A small damping resistor slows down the coil decay...

A small damping resistor speed up the coil decay...

The coil takes about 5uS after current shut off to spike and decay to zero

There is no absolute zero, you will see some very small voltage even after 200 uS. Also to decay in 5uS you have to build a really really good coil if it is possible at all.

[Waikiki_Sweep]

I'm sure I heard some talking of slight under damping sometimes being an advantage. Was it you Waikikii ? Anyway, there might be some tips to be found in HDPhilips two threads on his uno32 and his experiments with the QED hardware. He might have some scope shots posted showing how the damping behaves.

Yes. I saw it in on oscilloscope when underdamped coil plots some part of oscillation curve to counterweight decay voltage and move preamp from saturation in few microseconds earlier than properly damped coil.
It gives some advantages but it is unstable balance.

[Waikiki_Sweep]

One more reason for active damping:

PI detectors are relatively slow in comparison with VLF because 600 pps. is much less than 5 or 15 kHz.
So in real life you will see a situation when someone with PI machine searched the area than you can go after him with VLF and find a lot of items missed even if overage PI machine go deeper than VLF.
It happens because if coil moves fast detector loosing targets when it receives not enough samples per second to see that targets.

What to do? Increase frequency.
I would like to use PI with 2000-3000 pps. to get more staff with overage coil speed.
But it may not work because coil will accumulate eddy currents if there is not enough time to decay to zero between pulses.

That active damping you can see in Chance may help to build a faster PI detector. As it already works for short pulses series in Chance and grounding excess of decay current for short pulses.

[Tepco]

Nice discussion, I missed it (was off for a bit), will post more here, about Chance stuff, and two active damping patents mentioned.

[Tepco]

This circuit in Chance has nothing to do with active damping, or any damping at all. Damping is only with resistor. Purpose of another MOSFET can be seen from scope picture on author's website. Yellow trace is coil voltage, blue is current. During this burst pulse coil is alternatively charged thru one MOSFET, and short circuited with another in pause between pulses. So coil current is increased like stepped linear ramp, there is no flyback between pulses. Energy in coil is increased during one part of waveform, then stored during short circuit phase. Only at the end of burst pulse, both MOSFETS are open and then flyback happens. Later, another standard long pulse is generated, storing same amount of energy in coil, then releasing flyback. Soil mineralization and metal object will respond differently to two pulses, difference is used to establish ground rejection, as per original Minelab patent. This is the point, not active damping.

[Waikiki_Sweep]

This circuit in Chance has nothing to do with active damping, or any damping at all. Damping is only with resistor. Purpose of another MOSFET can be seen from scope picture on author's website. Yellow trace is coil voltage, blue is current. During this burst pulse coil is alternatively charged thru one MOSFET, and short circuited with another in pause between pulses. So coil current is increased like stepped linear ramp, there is no flyback between pulses. Energy in coil is increased during one part of waveform, then stored during short circuit phase. Only at the end of burst pulse, both MOSFETS are open and then flyback happens. Later, another standard long pulse is generated, storing same amount of energy in coil, then releasing flyback. Soil mineralization and metal object will respond differently to two pulses, difference is used to establish ground rejection, as per original Minelab patent. This is the point, not active damping.

Thank you to shed more light on how Chance works.

I have found one of charts on md4u.ru which showed coil voltage in malfunctioning Chance when shorting MOSFET is not working.
You can see that without shorting flyback starts after every short TX impulse if there is no shorting between impulses.
So properly working Chance just damping small flybacks from short impulses.

Chance schematic looks good to use in active damping too just another software have to be created.

Frankly I am little disappointed in Chance because of quality of discrimination. I can clearly hear difference between coins and iron junk but gold rings are at risk to be missed because they have same low tone as iron has. Use of discrimination modes will reduce sensitivity to gold signals too. Because I am looking for gold rings, not coins so I will have to dig all targets to be sure.

But probably already assembled Chance can be recycled by creating a new software which will have no discrimination but more pps. speed, as well as speed of reaction on targets and more sensitivity and active damping because it has shorting MOSFET already, e.t.c.
Chance is good platform for experimenting anyway.

[Silver Dollar]

Here is the code for the 16 pulses in Chance software. He opens or shorts the second FET before all the pulses then
reverses it after the pulses (b0 is TX FET and b1 is shorting FET);

L01C0:
cbi PORTB,b1
sbi PORTB,b0
ldi r16,k16
L01C3:
dec r16
brne L01C3
; ----- branch on last line
cbi PORTB,b0
ldi r16,k78
L01C7:
dec r16
brne L01C7
; ----- branch on last line
rjmp L01CA
; ----------- jump on last line
L01CA:
sbi PORTB,b0
ldi r16,k1A
L01CC:
dec r16
brne L01CC
; ----- branch on last line
cbi PORTB,b0
ldi r16,k74
L01D0:
dec r16
brne L01D0
; ----- branch on last line
rjmp L01D3
; ----------- jump on last line
L01D3:
sbi PORTB,b0
ldi r16,k1E
L01D5:
dec r16
brne L01D5
; ----- branch on last line
nop
cbi PORTB,b0
ldi r16,k70
L01DA:
dec r16
brne L01DA
; ----- branch on last line
nop
sbi PORTB,b0
ldi r16,k22
L01DF:
dec r16
brne L01DF
; ----- branch on last line
rjmp L01E2
; ----------- jump on last line
L01E2:
cbi PORTB,b0
ldi r16,k6C
L01E4:
dec r16
brne L01E4
; ----- branch on last line
sbi PORTB,b0
ldi r16,k27
L01E8:
dec r16
brne L01E8
; ----- branch on last line
cbi PORTB,b0
ldi r16,k67
L01EC:
dec r16
brne L01EC
; ----- branch on last line
rjmp L01EF
; ----------- jump on last line
L01EF:
sbi PORTB,b0
ldi r16,k2B
L01F1:
dec r16
brne L01F1
; ----- branch on last line
nop
cbi PORTB,b0
ldi r16,k63
L01F6:
dec r16
brne L01F6
; ----- branch on last line
sbi PORTB,b0
ldi r16,k30
L01FA:
dec r16
brne L01FA
; ----- branch on last line
cbi PORTB,b0
ldi r16,k5E
L01FE:
dec r16
brne L01FE
; ----- branch on last line
rjmp L0201
; ----------- jump on last line
L0201:
sbi PORTB,b0
ldi r16,k34
L0203:
dec r16
brne L0203
; ----- branch on last line
cbi PORTB,b0
ldi r16,k5A
L0207:
dec r16
brne L0207
; ----- branch on last line
nop
sbi PORTB,b0
ldi r16,k38
L020C:
dec r16
brne L020C
; ----- branch on last line
nop
cbi PORTB,b0
ldi r16,k56
L0211:
dec r16
brne L0211
; ----- branch on last line
nop
sbi PORTB,b0
ldi r16,k3C
L0216:
dec r16
brne L0216
; ----- branch on last line
cbi PORTB,b0
ldi r16,k52
L021A:
dec r16
brne L021A
; ----- branch on last line
rjmp L021D
; ----------- jump on last line
L021D:
sbi PORTB,b0
ldi r16,k3F
L021F:
dec r16
brne L021F
; ----- branch on last line
nop
cbi PORTB,b0
ldi r16,k4F
L0224:
dec r16
brne L0224
; ----- branch on last line
nop
sbi PORTB,b0
ldi r16,k42
L0229:
dec r16
brne L0229
; ----- branch on last line
cbi PORTB,b0
ldi r16,k4C
L022D:
dec r16
brne L022D
; ----- branch on last line
nop
sbi PORTB,b0
ldi r16,k44
L0232:
dec r16
brne L0232
; ----- branch on last line
nop
cbi PORTB,b0
ldi r16,k4A
L0237:
dec r16
brne L0237
; ----- branch on last line
nop
rcall L0296
rcall L0296
sbi PORTB,b0
sbi PORTB,b1
cbi PORTC,b5
ldi r16,k44

[Silver Dollar]

I agree having a PCB with everything there makes for a good platform to try some different code.
I want to get mine running in stock form then do a few tweaks and see what it can do.

[baum7154]

Response to Wakiki Sweep comments on BB SAILOR re-post

You are most welcome my friend!

I am flattered that you attribute to me the quotes of others I consider far more knowledgeable than I on the topic of metal detection.

I appreciate you correcting your original post from '---Quote (Originally by baum7154)' to 'Originally Posted by baum7154' but it would have been far more accurate to have said "Re-posted from an Original post by BB SAILOR (aka Joe Ragowski) with a link to a patent application on the topic of active MOSFET Damping.

I provided this re-post in hope that it would support your effort to make a slow coil act like a faster one through active coil damping and I eagerly await your progress on that front.

Regards,

Dan

[Tepco]

Forceful damping:

Using some form of active damping can be interesting idea, but not problem free. Two patents mentioned here propose using voltage controlled resistor or constant current sink instead of damping resistor. Everything mentioned in papers is absolutely correct from technical standpoint but as usual, some practicalities will tend to spoil these ideas. For purpose of this topic, speeding up fast coils, or compensating for added capacitance of coil shield, cable etc, this can produce very little, if any improvement, due to few simple reasons. In both patents, coil is considered as lumped element, inductance + some series resistance, having certain L\R ratio or time constant. If this model describe coil accurately in given situation, then discharging it with something other than resistor will have advantage, but in our situation this is not the case. With coil self-capacitance, added shield and cable capacitance circuit will act like LC resonant circuit (actually like more complex combination of distributed parameter circuits but simple LC can be used as a quite good approximation) having some specific self-resonant frequency. And this is the problem: fastest way of damping such circuit is purely resistive load. Circuit will refuse to “source” energy any faster, no matter how you try to sink it. This is why current sink, or time variable resistance wont help much.


Now, finally to add one suggestion how to actually achieve speeding up the coil electronically. From my point of view, practically only way is not passive or active damping, but something that can be called “forced damping”, or call it as you like. Idea is to inject another short pulse of opposite polarity from suitably high voltage source to forcefully remove charge from capacitive components faster. Obviously, another power source of voltage comparable to flyback, and another switch, presumably MOSFET, timed to produce sub-uS pulse. May look complicated but not need to be. Low power source is needed, cheap IC and few parts around. With good gate driver chip, MOSFET can be commanded fast enough, timing is very fast but this is not too problematic. No need to be precise or processor controlled to perfectly discharge circuit, just close, then resistor can finish. After all, we not trying to speed up coil 5 times, 50% can be enough, no need to be perfect. At some opportunity, when I again get close to my hardware, will try to assembly, or at least improvise something similar to see results, for now, just an idea open for discussion.

[Waikiki_Sweep]

I am sorry for inaccurate quotes. Forum software just put the "Originally Posted by baum7154" info header of every quote against my intentions. I hope everyone has understanding who and what telling because my post stays just after that block with bbsailor message.

[Waikiki_Sweep]

Now, finally to add one suggestion how to actually achieve speeding up the coil electronically. From my point of view, practically only way is not passive or active damping, but something that can be called “forced damping”, or call it as you like. Idea is to inject another short pulse of opposite polarity from suitably high voltage source to forcefully remove charge from capacitive components faster. Obviously, another power source of voltage comparable to flyback, and another switch, presumably MOSFET, timed to produce sub-uS pulse. May look complicated but not need to be. Low power source is needed, cheap IC and few parts around. With good gate driver chip, MOSFET can be commanded fast enough, timing is very fast but this is not too problematic. No need to be precise or processor controlled to perfectly discharge circuit, just close, then resistor can finish. After all, we not trying to speed up coil 5 times, 50% can be enough, no need to be perfect. At some opportunity, when I again get close to my hardware, will try to assembly, or at least improvise something similar to see results, for now, just an idea open for discussion.

It is interesting idea. Also if TX is a positive inpulse than blyback is negative (opposite direction). Can we use some very short positive TX at the moment of negative flyback to annihilate opposite voltages partially?

[baum7154]

Thank you to shed more light on how Chance works.

I have found one of charts on md4u.ru which showed coil voltage in malfunctioning Chance when shorting MOSFET is not working.
You can see that without shorting flyback starts after every short TX impulse if there is no shorting between impulses.
So properly working Chance just damping small flybacks from short impulses. [ATTACH]28999[/ATTACH]

Chance schematic looks good to use in active damping too just another software have to be created.

Frankly I am little disappointed in Chance because of quality of discrimination. I can clearly hear difference between coins and iron junk but gold rings are at risk to be missed because they have same low tone as iron has. Use of discrimination modes will reduce sensitivity to gold signals too. Because I am looking for gold rings, not coins so I will have to dig all targets to be sure.

But probably already assembled Chance can be recycled by creating a new software which will have no discrimination but more pps. speed, as well as speed of reaction on targets and more sensitivity and active damping because it has shorting MOSFET already, e.t.c.
Chance is good platform for experimenting anyway.

Are you running the V1.2.1 software?

[Waikiki_Sweep]

Using some form of active damping can be interesting idea, but not problem free. Two patents mentioned here propose using voltage controlled resistor or constant current sink instead of damping resistor. Everything mentioned in papers is absolutely correct from technical standpoint but as usual, some practicalities will tend to spoil these ideas. For purpose of this topic, speeding up fast coils, or compensating for added capacitance of coil shield, cable etc, this can produce very little, if any improvement, due to few simple reasons. In both patents, coil is considered as lumped element, inductance + some series resistance, having certain L\R ratio or time constant. If this model describe coil accurately in given situation, then discharging it with something other than resistor will have advantage, but in our situation this is not the case. With coil self-capacitance, added shield and cable capacitance circuit will act like LC resonant circuit (actually like more complex combination of distributed parameter circuits but simple LC can be used as a quite good approximation) having some specific self-resonant frequency. And this is the problem: fastest way of damping such circuit is purely resistive load. Circuit will refuse to “source” energy any faster, no matter how you try to sink it. This is why current sink, or time variable resistance wont help much.

It is true, slow coil may release energy slow.

But with using different damping resistors we can see a different decay curves.

Idea is to bring coil voltage into preamp input range with low resistance load as fast as we can than use higher dumping resistance to read samples.

Probably we just need to connect additional damping resistor for a short moment before first sample.

How about 1k dumping resistor and 100 ohm resistor in line with Shottky diode.
When flyback is more than 200mV diode is open and damping is 100 Ohm. Once it goes below 200mV diode will close and damping will stay 1 kOhm.