damping resistor

[baum7154]

Using my scope connected to op amp output today the best I could tell ringing/decay of coil and feed in circuit was 953 kHz. Using that number and running the formula in the first post of this thread I come up with 1003 ohms. Setting my adjustable resistor network for that value the coil will not operate at Guard Interval set at 10. Network must be adjusted to 1020 ohms before detector will run at GI =10. I check discrimination for various types of metals and they are fairly accurate...gold is good, iron good, lead good, copper good, aluminum good, nickel good above 6 units of signal strength. I am now trying a range of values to see if sensitivity improves without degrading discrimination. Have gone as far as 1150 ohms but results are still unclear.

I had 1260 ohms in circuit last week and noticed that discrimination suffered, most everything wanted to show as gold/aluminum.

I did do resonance on just the coil and feed line months ago using BB Sailor's technique with a signal gen and scope. I believe it resonated at about 1.045 mhz but that was out of circuit. The formula for this resonance on this 335uh coil puts the damping at 1100 ohms. Looks like the in circuit resonance is about 9% lower.

I'll post results here when they are finalized.

Thanks for the help and I welcome any further suggestions.

Regards

Dan

[Tepco]

Results are interesting. I mentioned measurement of coil SRF only, mostly because comparison between shielded and unshielded coil is interesting, to see how much shield will affect performance, also to devise good shielding method. These results, 335uH coil and 1.045MHz SRF, indicate only some 69pF self capacitance, very good indeed. But 953kHz measured with detector circuit indicate some 83pF, this is bit too optimistic to expect only 14pF added by detector ?!? I suspect initial coil alone measurement was somewhat higher, but even then I expected greater change. After all, with complete circuit only minimal sampling time is important, SRF and capacitance is just parameter for comparison.

[Tinkerer]

I have not seen any mention about the Joules stored in the coil affecting the damping. If I remember right, this has a very significant influence of how fast a PI coil can be damped.

[Tepco]

Not in this thread, but I mentioned that at least dozen times before...

[baum7154]

I have not seen any mention about the Joules stored in the coil affecting the damping. If I remember right, this has a very significant influence of how fast a PI coil can be damped.

My detector does have the ability to drop supply voltage in .7 volt steps to 72% of supply voltage and dropping voltage does have a dramatic effect on guard time on slower coils. Perhaps this could be a useful in a test.

[green]

I have not seen any mention about the Joules stored in the coil affecting the damping. If I remember right, this has a very significant influence of how fast a PI coil can be damped.

I'm still trying to learn. Would appreciate being corrected where I'm wrong. Rd for critical damping doesn't change with stored energy. With spice I see a small increase in time to damp to the same volt level if I double the current. The hardware circuit I'm using clamps the fly back at 450 volts. Increasing the current increases the clamp time which adds to the total time, with the decay time staying the same. Maybe other circuits act different. Including spice simulation showing effect of changing Rd for critical damping and under damped. The first plot the amplifier Fc is above coil resonance which measured 900 khz with Rd disconnected. Rd for critical damping equals pi * l * Fr = 850 ohms. The second plot I tried to simulate the CHANCE amplifier. Divided OP37 BW by gain of 470 = 170 khz. Shunted 470k fdbk resistor with a capacitor for same response. Maybe not correct. Playing with the circuits in spice has been fun. Making changes and measurements is a lot quicker. Don't know how good spice works for non inverting input simulation.

[baum7154]

[Regarding measuring the resonance across the input diodes I think the capacitance of the scope probes would interfere with the measurement so I think it is better to go to the output of the op amp to isolate the probe from the coil. Does that make sense?]

I think you will get the same answer either location. Measuring resonance in circuit includes the added driver capacitance and should measure lower than coil resonance. I think this is the frequency you are trying to damp. Is the resonance what you were trying for when you wound the coil? You still adjust damping for best response. If you can't adjust the delay to the minimum you want, knowing the resonance might explain why.

I agree that the in circuit resonance is the damping I seek. When I wound the coil I was not shooting for a particular Fr, just the lowest capacitance attainable. The Fr resulting was a direct result of the effort for low capacitance. I'm thinking I'll re-run the Fr measurement again to insure any error is minimized. I need to borrow a signal gen from work to do it again. I have always been able to get this coil to function at a Guard Interval setting of '10' which is the lowest delay that CHANCE PI allows.

Dan

[Tinkerer]

I'm still trying to learn. Would appreciate being corrected where I'm wrong. Rd for critical damping doesn't change with stored energy. With spice I see a small increase in time to damp to the same volt level if I double the current. The hardware circuit I'm using clamps the fly back at 450 volts. Increasing the current increases the clamp time which adds to the total time, with the decay time staying the same. Maybe other circuits act different. Including spice simulation showing effect of changing Rd for critical damping and under damped. The first plot the amplifier Fc is above coil resonance which measured 900 khz with Rd disconnected. Rd for critical damping equals pi * l * Fr = 850 ohms. The second plot I tried to simulate the CHANCE amplifier. Divided OP37 BW by gain of 470 = 170 khz. Shunted 470k fdbk resistor with a capacitor for same response. Maybe not correct. Playing with the circuits in spice has been fun. Making changes and measurements is a lot quicker. Don't know how good spice works for non inverting input simulation.

We are all learning. The whole purpose of the discussions is to learn something.
Spice simulation is very interesting, but only as good as the input we give it. By the way, it would help if you would add the LTSpice file to make it easier for us to run your sim. I do not find your model of the Mosfet.
I found that the modeling of the coil gives different results with different input. The inductance model has an input for parallel capacitance, series and parallel resistance etc. Adding a capacitance separately makes for a different result.
With real coils we found that the added capacitance of the shield somehow does not give the expected results. One tentative explanation is that some of the inter-wire capacitance and shield capacitance etc. may act in series and so change the expected result.

We also found in real life that it is good to over damp a little for the sake of stability.

The 2p feedback capacitor is good for stretching the 10us TC target response a little.

[green]

[ it would help if you would add the LTSpice file to make it easier for us to run your sim. I do not find your model of the Mosfet.]

Would be happy to if someone would tell me how. The mosfet is near the bottom of the list between five Fairchild fets in my LT spice program. I tried changing to internal capacitance and got the same result. Been using external capacitor so some could draw circuit in LT spice to play.

[bbsailor]

Green, Tinkerer and all interested

The circuit you are analyzing is a mono coil where the same coil switches between being the TX coil and then quickly becomes the RX coil however there are three variables to consider in your analysis.

Flyback voltage is a function of the length of the TX pulse width with longer pulses causing the current to rise higher on the current growth curve based on the coil time constant (TC) calculated by the coil inductance being divided by the coil resistance. A 300uH coil that is 5 ohms (total, including Mosfet on-resistance, coax resistance and coil wire resistance) will have a TC of 60us so that in 1TC the current grows to about 63 percent of the maximum current and at 120us TX width be 2TCs or about 85 percent of the maximum current and so on until 5 TCs where the current grows to about 99 percent of max. Higher current means a higher flyback voltage and more energy that is needed to be damped. Run your coil design (inductance, resistance and voltage applied) in program MiscEL to see the amplitude of the flyback voltage.

An optimumly damped coil is only valid for a specific TX pulse width. However many PI designs, such as Eric Fosters PI designs, have about a 10 percent variable TX frequency to allow the user to find a frequency or harmonic that is not in sync with local noise or related power frequency harmonics (50Hz in Europe or 60Hz in the USA). When you widely vary the TX frequency (from hundreds to thousands of pulses per second) you need a variable damping resistance to cover the full range of flyback voltages caused by the widely varying TX frequency range. Eric Foster's PI designs used the 3000PPS (pulses per second) rate with about a 10 percent variation around 3000PPS. Eric recommends that a slight over-damping be used or about 10 percent less resistance than optimum calculated damping.

Here are the three variables to keep in mind when calculating or measuring optimum damping resistor values.

1. The coil driver MOSFET clamping voltage rating should exceed the actual flyback voltage to obtain the fastest sampling.
2. The mono coil is subject to the input resistor (usually 1K ohm) being in parallel with the damping resistor value when the clamping diodes are conducting above about 0.7V.
3. Once the flyback voltage falls below 0.7V the input resistor is no longer in parallel with the damping resistor value.
Once you attempt to enter the fast coil design area (at, near or below 10us delay) you need to consider how long it takes the first amplifier stage to come out of saturation. That is why 2 stages of 33X gain is better than one stage of 1000X gain from a potential sampling speed persepective.

A 10 percent variable damping resistor total value using a variable pot in parallel with a higher wattage fixed damping resistor will allow you to fine tune your PI to the TX pulse width. Most power will be dissipated through the fixed damping resistor with only about one tenth the power being on the variable pot so high voltage arcing in the pot should not be a problem.

Keep these variables in mind when running your damping resistor models.

I hope this helps?

bbsailor

[green]

[Run your coil design (inductance, resistance and voltage applied) in program MiscEL to see the amplitude of the flyback voltage.]
Do you think LT spice doesn't give a valid simulation? I haven't figured out how to use MisEL yet.

[An optimumly damped coil is only valid for a specific TX pulse width.]
Including a spice simulation at two current levels. The decay looks the same, trying to understand why Rd needs to be adjusted.

[bbsailor]

[Run your coil design (inductance, resistance and voltage applied) in program MiscEL to see the amplitude of the flyback voltage.]
Do you think LT spice doesn't give a valid simulation? I haven't figured out how to use MisEL yet.

[An optimumly damped coil is only valid for a specific TX pulse width.]
Including a spice simulation at two current levels. The decay looks the same, trying to understand why Rd needs to be adjusted.

Green,

Download the latest version of MiscEl. Go to the calculations tab and then to the "inductor charge" calculator. Enter the voltage (U), inductance in microhenries (u) and resistance of the coil. See the graph and put your cursor on the graph at the TX pulse width of your design. Note the current below the graph.

Now, go to the "inductor discharge" calculator and enter the noted current from the previous screen. See the flyback voltage in the data below the graph.

If the damping resistor(Rd) is 1000 ohms and Rin (the input resistor to the first amp stage) is 1000 ohms then Rd is 500 ohms until the flyback voltage drops below 0.7V. This reflects the shape of the discharge time constant which should be 5 times faster than the TC of the target that you are seeking for optimal stimulation of that target.


Simulators are only as good as the accuracy of the input data.

The most accurate way to optimize damping is to attach an oscilloscope to the output of the first PI amplifier stage and adjust the Rd value until the earliest sampling time is seen on the discharge curve. This curve takes into account all the circuit loading on the TX/RX coil. The best way to speed up the coil is to use the lowest capacitance coax cable, near or below 20pf per foot, the lowest dielectric constant wire insulation (like Teflon) and a spacer between the coil wire bundle and the shield. Shorter TX pulses can be damped more easily than longer TX pulses. It is all a tradeoff between the target TC that you are seeking and the design of a coil to focus on those target TCs. One coil is not optimal for both targets like coins and small gold nuggets. For small gold nuggets you need a smaller coil in the 3" to 5" diameter range with very short TX pulses, less than 50 microseconds. Larger coils work well on the beach with 100uS to 200uS TX pulses and a delay of about 15uS for the wet beach zone and 10uS for the upper beach or dry beach zone.

Remember, the value of Rd defines the coil discharge TC and is very closely related to the smallest size (TC) of the targets you can detect.

Read the help screens in MiscEL. This is a pretty good calculator to help you form a better mental model of your coil design and related TX pulse and current parameters.

This should help!

bbsailor

[Tinkerer]

[Run your coil design (inductance, resistance and voltage applied) in program MiscEL to see the amplitude of the flyback voltage.]
Do you think LT spice doesn't give a valid simulation? I haven't figured out how to use MisEL yet.

[An optimumly damped coil is only valid for a specific TX pulse width.]
Including a spice simulation at two current levels. The decay looks the same, trying to understand why Rd needs to be adjusted.

The LTSpice simulations are an excellent working tool. However, There are so many variables in our PI circuits and each variable influences the results. I hope in time to have models with accounting for all the variables, but it may take a long time.

The MiscEl app is not updated for the latest Windows, so it starts with a lot of error messages that you have to click on, but then it works just fine. It is an excellent tool.
Below is a LTSpice file, zipped. It only uploads when zipped. This way we can use each others simulations without having to redraw the whole circuit.

[Tinkerer]

Hi bbsailor,

nice to see you again. Thanks for your excellent posts that are really very helpful.

All the best
Tinkerer

[Tepco]

Hope this can be interesting to someone, instead of simulation I made few measurements. Original Chance coil is used, spider design, nominal 400uH, measured 397uH, made from 0,7mm solid wire on 4mm plexiglass former. I tested SRF, (coil alone, and with cable and tween lead) and minimal sampling speed and this is what I get:

Coil alone SRF is very high, 1.46MHz, indicating only 29pF self capacitance. Connected to 1.8m RG 58 cable it drops to 595kHz, now total capacitance is 182pF. With 93pF\m cable capacitance should be 167pF, I measured it to 172pF using resonant LC meter, close enough, but simplified calculation, just adding coil self capacitance, require only138pF. With tween lead SRF is 835kHz or 92pF in total, but I measured cable alone to 72pF (this may be questionable due to nature of instrument used), 63pF is required with simple calc. Probe capacitance is very small (some 3.5pF x10 probe), I ignored it.


Now, even best part. This is connected to TX stage, working at 100uS pulse at 1kHz, with IRF740 switch, and observed using 2 stage preamp, known to be fast enough, tested well below 10uS before. Using RG58 cable and original 390R damping I get nicely damped waveform at around 17uS, but interestingly, increasing this value to over 1K waveform remained about same, at very high value (1020R) some very fast ringing appear. Without Rdamp coil ringdown is 310kHz (!) indicating unrealistic 660pF capacitance (!) Fast diode in circuit not changed waveform at all when added. Adding very small (ten's of pF or less) capacitance at TX end produce underdamped waveform and increase speed to less than 14uS. Same setup using tween lead cable produce maybe 1uS faster waveform, diode again failed to produce any effect, now ringdown without Rdamp is around 500kHz, indicating more realistic but again too high 250pF total circuit capacitance.


Conclusion: forget simulations, build it and measure! Number of factors are involved, coil is not simple lumped LC circuit but distributed parameter , cable too, TX stage behavior is highly nonlinear (simulation may account for this to some degree), coil own eddy current decay counts, material used and geometry...Not that simple, setting up complete model to account for all this is very messy business.