Minipulse Plus Rev My time has come!

[mushaba]

Another factor that governs the coil discharge is the gate driving circuit of the MOSFET. Using active turn-off (as opposed to passive) can dramatically boost the flyback voltage, and hence the speed of turn-off. In a circuit I was experimenting with recently, I was able to increase the value of the damping resistor after implementing active turn-off. Of course you have to be careful not to push the MOSFET into avalanche mode.

I did a quick search the paper you mentioned, but couldn't find it either.

Taken from a study os some Chinese dudes.. https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=8409349


Conclusion : In order to solve the problem of voltage spikes and oscillations caused by the high frequency switching of MOSFETs in power electronic converters, the NAGD is proposed to suppress voltage spikes and oscillations in the turn-off process. The simulation and experimental results show that under the threshold of the gate voltage, the higher the gate voltage is, the better the suppression effect of voltage spike and oscillation will be. At the same time, for higher parasitic inductance and bus voltage, not only can voltage spike and oscillation be effectively suppressed, EMI at high frequencies can also be effectively suppressed. In addition, the control of the NAGD is simple and requires fewer additional components, which is beneficial to the realization of the circuit.

[bbsailor]

Another factor that governs the coil discharge is the gate driving circuit of the MOSFET. Using active turn-off (as opposed to passive) can dramatically boost the flyback voltage, and hence the speed of turn-off. In a circuit I was experimenting with recently, I was able to increase the value of the damping resistor after implementing active turn-off. Of course you have to be careful not to push the MOSFET into avalanche mode.

I did a quick search the paper you mentioned, but couldn't find it either.

Qiaozhi,

Please tell us what the higher value of the damping resistor was in your experiment compared to the original value? I would assume that using active turn off eliminates the COSS of the MOSFET and other sources of capacitance and loading as seen by the flyback pulse?

Reducing capacitance as seen by the flyback pulse reduces the energy in the oscillations that require an Rd value to quickly get to the point where the RX circuit can be turned on, otherwise known as delay time.

Since the Rd value determines the coil discharge slope, this becomes very critical for those seeking targets with less than a 2 micro second target decay.

The optimization or TX And RX pulse parameters comes from reverse engineering the PI parameters based on the desired targets time constant.

Do you agree with my above observation?

Joseph J. Rogowski

[bbsailor]

Another factor that governs the coil discharge is the gate driving circuit of the MOSFET. Using active turn-off (as opposed to passive) can dramatically boost the flyback voltage, and hence the speed of turn-off. In a circuit I was experimenting with recently, I was able to increase the value of the damping resistor after implementing active turn-off. Of course you have to be careful not to push the MOSFET into avalanche mode.

I did a quick search the paper you mentioned, but couldn't find it either.

Qiaozhi,

Please tell us what the higher value of the damping resistor was in your experiment compared to the original value? I would assume that using active turn off eliminates the COSS of the MOSFET and other sources of capacitance and loading as seen by the flyback pulse?

Reducing capacitance as seen by the flyback pulse reduces the energy in the oscillations that require an Rd value to quickly get to the point where the RX circuit can be turned on, otherwise known as delay time.

Since the Rd value determines the coil discharge slope, this becomes very critical for those seeking targets with less than a 2 micro second target decay.

The optimization of TX and RX pulse parameters comes from reverse engineering the PI parameters based on the desired targets time constant.

Do you agree with my above observation?

Joseph J. Rogowski

[Qiaozhi]

Qiaozhi,

A paper that Eric Foster posted (that I can not find) mentioned that the coil discharge Time Constant (TC) should be 5 times faster than the desired target TC to fully stimulate it. The paper stated that turning off the coil discharge TC any faster did not stimulate the target any more than doing it 5 times faster than the target TC. I am just adding this to the discussion so as to fully address this little spoken of variable.

The coil discharge TC is governed by how vertical the discharge spike is. More vertical is faster. This discharge spike slope is governed by the coil inductance divided by the damping resistor (Rd) value and any other resistance that is in parallel with Rd such as, in a mono coil, the series resistor to the op amp plus the clamping diodes that parallel the Rd value with the op amp input resistor value while the diodes are conducting. All mono coil discharge slopes have a kink in the discharge slope at the point until the clamping diodes open and then resumes a more vertical discharge slope.

If we are doing target and pulse characteristic analysis we need to assume that the desired target is being fully stimulated in order to calculate any other variables.

What is your opinion on this?

Joseph J. Rogowski

Another factor that governs the coil discharge is the gate driving circuit of the MOSFET. Using active turn-off (as opposed to passive) can dramatically boost the flyback voltage, and hence the speed of turn-off. In a circuit I was experimenting with recently, I was able to increase the value of the damping resistor after implementing active turn-off. Of course you have to be careful not to push the MOSFET into avalanche mode.

I did a quick search the paper you mentioned, but couldn't find it either.

[bbsailor]

As you said, everything is a trade-off. A lot depends on what you're searching for. If it's small gold nuggets, then you may want to increase the TX pulse rate, and use a series resistor to limit the maximum current in order to achieve flat-topping, but (as a consequence) lose a lot of depth. In that case it's probably unavoidable with the standard PI setup. For coin searching, cutting off the current before it flat-tops may be the best solution if you don't want to lower the TX pulse rate. For instance, try 1000pps with a 100us pulse width and a 3R3 series resistor. Then do the same test (same pps and pulse width) but with a suitable series resistor to allow the current to flat-top. The former test will have the best result on coins, as the magnetic field will be much greater, and it more than compensates for the opposing field from any eddy currents in the target.

Qiaozhi,

A paper that Eric Foster posted (that I can not find) mentioned that the coil discharge Time Constant (TC) should be 5 times faster than the desired target TC to fully stimulate it. The paper stated that turning off the coil discharge TC any faster did not stimulate the target any more than doing it 5 times faster than the target TC. I am just adding this to the discussion so as to fully address this little spoken of variable.

The coil discharge TC is governed by how vertical the discharge spike is. More vertical is faster. This discharge spike slope is governed by the coil inductance divided by the damping resistor (Rd) value and any other resistance that is in parallel with Rd such as, in a mono coil, the series resistor to the op amp plus the clamping diodes that parallel the Rd value with the op amp input resistor value while the diodes are conducting. All mono coil discharge slopes have a kink in the discharge slope at the point until the clamping diodes open and then resumes a more vertical discharge slope.

If we are doing target and pulse characteristic analysis we need to assume that the desired target is being fully stimulated in order to calculate any other variables.

What is your opinion on this?

Joseph J. Rogowski

[mushaba]

Thanks a lot guys the amount of information is huge and precious,please give me some time as a learning curve..My MPP does work but I want to get the most out of it,it's already has enough detection depth even with my erroneous coils..I see the hole build more as a new field lesson for me and at the end of this course I will have a working detector that I understand it's working at it's most part.I know I won't get it all at once but I'm a good learner,still in a field I've never been involved with..I work with Atmel mcus,I have my own old stk500 but it does it's job perfectly for my needs so far,I work most project of mine with the mcu assembly and for more complex designs I use Bascom.Happy to help if I can..I must admit when I started involving with mds the hammerhead (5 version) caught my eye because it was a complete project with lot of exlpanation and lot of flexibility...A friend guided me to Svens Mirage which I think that is also a great project,loved George's Crossbow and ended up with MPP due it's two stage receiver..Also have a Delta Pulse Legend that I don't think I will work with because no I structions are available.The gift for me is knowledge,and I thank you all responding or no for all the usefull things you post here in geotech..Now days there are lots of good MD with a price range affiortable but as Steven Tayler signs in Amazing "life is a journey not a destination"...Just today I received my ptfe silverplatted wire and I'm anxious to go to my homeshop and get down to it...Stay tuned and sorry for writing too many of my thoughts and also sorry for my bad English,I think hundrends of braincells are destroyed when you are reading my posts...

[Qiaozhi]

This is one of the design trade-offs in PI detectors. We want a high pulse rate for better response and sensitivity but this then requires a short pulse time with limits the coil TC if we want a 'flat-top' coil current. Or with a longer pulse time we must decrease the pulse rate to be able to do the sampling.

This then gets back to the do we need a 'flat-top' current? I have studied the SD2000 schematic and timing. They do two TX pulse widths with the same coil series resistance.
Then have short and long sampling times. I do not see how they get a 'flat-top' coil current with the short TX on times so maybe it isn't required.
But seems that matching the pulse time to Target's TC is more important.

As you said, everything is a trade-off. A lot depends on what you're searching for. If it's small gold nuggets, then you may want to increase the TX pulse rate, and use a series resistor to limit the maximum current in order to achieve flat-topping, but (as a consequence) lose a lot of depth. In that case it's probably unavoidable with the standard PI setup. For coin searching, cutting off the current before it flat-tops may be the best solution if you don't want to lower the TX pulse rate. For instance, try 1000pps with a 100us pulse width and a 3R3 series resistor. Then do the same test (same pps and pulse width) but with a suitable series resistor to allow the current to flat-top. The former test will have the best result on coins, as the magnetic field will be much greater, and it more than compensates for the opposing field from any eddy currents in the target.

[waltr]

Hi waltr,
What you have done with the HH is awesome I think im going to have to learn enough code to adjust a pi, I have a HH just sitting there done and waiting.
Be Ready for lots of questions. LOL
Anyway on coil tc and charging, So if I have a 450 uh coil at 1.9 ohm =a TC of 236.84 for this coil and we want on time of three to five times that.
Which would be 710.52 us which is huge so we take a resistor with a value of 13.5 ohm add it in line and now the on time would be 99.48 us.??
Is that correct that would be the way to get a 450 uh coil to peak with a lower on time?
Thanks

Thanks and please ask questions. My HH2 thread is open for anyone to ask and discuss, that is how we learn.
I and others have some threads in the Programming sub-forum in hopes of help people move to using micro-controller/processor instead of timing chips.
Ask there on those topics.

In general, yes, adding series R to the coil is a way to decrease the coil TC and get the current to 'flat top' before turning off.
This has been under lots of discussion and with some disagreement. George (Qiaozhi) posted a link above to one of these discussions.
What is more important is experimenting to see what works best.

No, I meant the TX pulse rate. If the pulse rate is too high, then there may not be enough time to allow the current to flat-top, and still leave enough time to do everything else.

Ok but I see this as duty cycle: TX Pulse time verse TX pulse rate.
If you keep the pulse rate constant and increase pulse time then TX ON duty cycle increases leaving less time to sample signals.

This is one of the design trade-offs in PI detectors. We want a high pulse rate for better response and sensitivity but this then requires a short pulse time with limits the coil TC if we want a 'flat-top' coil current. Or with a longer pulse time we must decrease the pulse rate to be able to do the sampling.

This then gets back to the do we need a 'flat-top' current? I have studied the SD2000 schematic and timing. They do two TX pulse widths with the same coil series resistance.
Then have short and long sampling times. I do not see how they get a 'flat-top' coil current with the short TX on times so maybe it isn't required.
But seems that matching the pulse time to Target's TC is more important.

[Qiaozhi]

You say TX Rate but do you actually mean TX pulse width (with TX rate a side issue of having enough time between TX pulses, duty cycle, to measure decay)?

No, I meant the TX pulse rate. If the pulse rate is too high, then there may not be enough time to allow the current to flat-top, and still leave enough time to do everything else.

[godigit1]

Hi Mushaba,
I had a little time to mess with my new coil, and indeed i was able to recreate the flat topped coil current. I wasnt saying Green or anyone was wrong I just had not seen that on my fet before.
But with a Mpp rev d (which only has the 740 fet) at stock settings 1000pps i had to push my coil voltage above 460v before it became that pronounced.
Im goint to have read the thread posted by Qiaozhi and thank you Green for posting this as well it would have just flown by me without seeing your waveform.
Thank you for posting this thread Mushaba its also helped me put a couple of the puzzle peices together or at least i think I starting to get something through my thick skull.

Hi waltr,
What you have done with the HH is awesome I think im going to have to learn enough code to adjust a pi, I have a HH just sitting there done and waiting.
Be Ready for lots of questions. LOL
Anyway on coil tc and charging, So if I have a 450 uh coil at 1.9 ohm =a TC of 236.84 for this coil and we want on time of three to five times that.
Which would be 710.52 us which is huge so we take a resistor with a value of 13.5 ohm add it in line and now the on time would be 99.48 us.??
Is that correct that would be the way to get a 450 uh coil to peak with a lower on time?
Thanks

[waltr]

Here's a thread where we discussed flat-topping the coil current, and a simulation showing the benefits of allowing the coil current to top out -> https://www.geotech1.com/forums/show...top#post221739

However, this is not the whole story, as it's not always possible to flat-top the coil current due to the TX pulse rate being used. In order to top out the current, you will most likely have to lower the TX rate to allow enough time for this to occur. If you must use a higher pulse rate, then flat-topping may not even be possible. As usual, it's a bit of a trade-off, depending largely on what you're trying to achieve.

Thanks for the link to that discussion George. Good review.
I was trying to keep some basics simple for mushaba so he can get his detector actually working first.
I do believe he had the MOSFET going into Avalanche and causes issues. A series R helped there by reducing the total magnetic field in the coil.
He now has a lot of additional stuff to think about and try.
I try to keep with first order effects in simple explanations. I know there are many more orders that increases the complexity and can make the mind spin.

You say TX Rate but do you actually mean TX pulse width (with TX rate a side issue of having enough time between TX pulses, duty cycle, to measure decay)?

mushaba,
This is part of why I suggested measuring coil current and trying different series R values.
Also check response to various targets with different series R's.

If you like experimenting I suggest get and build a second board.
Get the first one housed and one a full working detector to use to find great things. Then the second board one for experimenting.
I have one full working detector and three other boards I use for experimenting. When I find a circuit, or code (I use a processor for all timing), change that makes a good improvement I then mod the working detector.

[Qiaozhi]

Here's a thread where we discussed flat-topping the coil current, and a simulation showing the benefits of allowing the coil current to top out -> https://www.geotech1.com/forums/show...top#post221739

However, this is not the whole story, as it's not always possible to flat-top the coil current due to the TX pulse rate being used. In order to top out the current, you will most likely have to lower the TX rate to allow enough time for this to occur. If you must use a higher pulse rate, then flat-topping may not even be possible. As usual, it's a bit of a trade-off, depending largely on what you're trying to achieve.

[mushaba]

All True Green but I was keeping explanations simpler to help mushaba get up to speed on the basics of PI design and so he can get this detector usable.

The think I have in mind after your and greens intell is at first we should have a coil with correct specs so it can "charge" fully.At it's discharge(I think "collapse" is the right word) the rapid discharge is creating the eddys we re looking for..constant (tx pulse)rate or constant (coil) current are the things to be investigated by me in the short future.
What "kind" of opamp setup is the receiver of the MPP?

[waltr]

Not saying if constant rate or constant current at Tx off is better or worse. I get a higher Rx signal with constant rate if average coil current is the same, and a higher Rx signal with constant current if peak current is the same. With the same average current, constant current peak is half the constant rate peak. Should stabilize sooner but I think I still get little more detection distance with constant rate if average current is the same. Just saying what I get, use what works best for you.

I have coil current in a control loop so when I compare different coils the Tx profile is always the same. Haven't tried comparing charging to 3 time constants with cc or cr.

All True Green but I was keeping explanations simpler to help mushaba get up to speed on the basics of PI design and so he can get this detector usable.

[green]

Yes, what we want is the coil current be be steady before turning off coil.
This then has the target exposed to a steady magnetic field and eddy current from increasing magnetic field have also settled to zero.
Now when TX is switched Off the magnetic field collapses very quick inducing the target to produce strong repulsing eddy currents.
It is the target's repulsing eddy current that affect the coil decay curve and are detected.

3 times is a minimum, some literature says 5 times but 3 to 5 times is not much change so 3 works well.

Try different series resistor values when you get to measuring TX on coil current with your scope.
With small series R the current will still be increasing when TX switches off but will level out with higher R.

Not saying if constant rate or constant current at Tx off is better or worse. I get a higher Rx signal with constant rate if average coil current is the same, and a higher Rx signal with constant current if peak current is the same. With the same average current, constant current peak is half the constant rate peak. Should stabilize sooner but I think I still get little more detection distance with constant rate if average current is the same. Just saying what I get, use what works best for you.

I have coil current in a control loop so when I compare different coils the Tx profile is always the same. Haven't tried comparing charging to 3 time constants with cc or cr.