Tweet
Hello JC1,
The method I mentioned has disadvantages and advantages. I spent some time converting from the java Circuit Simulator to LTSpice. Here are the results:
The nugget R & L values are from my calculations for a nugget that has resistance of near lead. The "Typcial 18 usec Tperiod" is a typical pulse where the coil charge time is 7.5 times longer than the collapsing field time. As you can see, the typical pulse method gives slightly better results, but this assumes you can recycle the energy as efficient as my method.
Equal power was used for all the above examples. Also I considered the fact that 4 times as many gathered samples per second yields twice the SNR. So the above results were resulting signal from nugget multiplied times square-root of sample rate. Also, if the pulse was half the time, then I made sure the coils voltage was doubled so as to maintain the same resulting peak current on the coil. The total energy put into the coil is equal to I^2. Also if the sample rate was 4 times greater then I cut the coils voltage by half. This is because four times as many pulses occur per second. Cutting the coils voltage in half equals to one-fourth the energy thereby making it equal in power dissipation.
Here are some reasons to use my square symmetrical pulse method:
1. To work with one voltage level; e.g. 6 volts. Rather than say 12 volts for charge and 400 volts during collapse time. Or in my case it would be more like 180 mV charge and 6 V during collapse time since I'm using a 1-turn coil.
2. To simply the design. I'm not an expert circuit designer and struggle with gotchas of circuitry that you pros know about. Well, they seem like gotchas to me, lol. It is very important to recycle the collapsing magnetic fields energy. So if I went with the typical design then I would have to charge the coil at say 180 mV and then recapture the collapsing field into a 6 volt source. I could use a 180 mV battery to charge the coil, lol, and then collapse into a capacitor that's charged to 6 volts and then recycle a small amount of the caps energy back into the 180 mV battery. Or I could do the reverse and use a 6 V battery and somehow efficiently maintain ~180 mV on a cap and then use that cap to charge to coil and then collapse the field directly into the 6 V battery. Either way you are going to lose energy in the energy conversion process. Perhaps an efficient method is to have a dozen 180 mV batteries. I could use one of the batteries to charge the coil and then collapse the field into all 12 batteries. Then every so often I would have to rotate the batteries. Are there any good 180 mV batteries? In my case 600 mV sound better anyway. Perhaps another method is using a tiny 600 mV battery and a large 6 V battery. The tiny 600 mV would charge the coil and the collapsing field would go into the 6 V. Then perhaps every minute or so the circuit would automatically charge a toroid and dump its energy into the 600 mV battery. This could perhaps achieve 90% efficiency. Last year I was working on something else and achieved 85% efficiencies, but efficiency was not my goal. Another option is using a transformer, but at a glance I don't care for that idea.
3. The present design seems almost as efficient as the typical method except I do not have to waist a lot of energy working with wide voltage sources and energy conversions. I can simply charge the coil with the 6 volt battery and then collapse the field back into the same 6 volt battery. According to my calculations, "my 18 usec Tperiod" yields equally efficient results all the way up to nuggets that are a few inches in diameter. A 10" nugget is over half as efficient as the typical method. Although, according to my calculations, I'll be able to detect a 10" nugget farther than I'd ever want dig. ... 10" nuggets, lol.
A few designs ago I had a simple fix for the big nugget issue. The fix was simply shorting the coil in-between coil charge and collapsing field periods for several usec's. This allowed the nuggets current to significantly collapse during this "wait period". In other words, a voltage is applied to the coil. So now there's current in the coil. Then the voltage is removed and replace with a short across the coil. This short maintains the coils current. So if the 1-turn coil is 2 uH and the transistors resistance is 0.01 ohms, then after 10 usec the coil's current drops 5%, or after 20 usec a 10% loss. That's not too bad. Note that the nuggets current will collapse during this coil shorted period because the coils di/dt is extremely low. Then the coil short is removed and the collapsing field is directed across the same 6 V battery.
Perhaps I'll put the "coil short" feature back in if the detector doesn't scan as deep as desired, but that will yield a 5% to 10% loss. Not sure if it will pay off. For now, it seems this method is better for nuggets smaller than 1", unless you use a really efficient method of working with wide voltage sources and energy conversions. You could use the dozen-battery method and rotate the batteries, but that doesn't sound like fun. Otherwise I'm wondering just how much loss there would be. Perhaps 20% to 40%? If you know of a method that's over 90% efficient then I would consider using the typical pulse method. Where there's a will there's a way.
Peace,
Paul
The method I mentioned has disadvantages and advantages. I spent some time converting from the java Circuit Simulator to LTSpice. Here are the results:
Code:
No losses are included. Nugget diameter: 0.1" (2 nH, 3200 uOhm) My 5 usec Tperiod: 920.75 mV Typical 18 usec Tperiod: 953.60 mV ----------- 1" My 5 usec Tperiod: 10.08 mV Typical 18 usec Tperiod: 12.92 mV ----------- 10" My 5 usec Tperiod: 1.93 uV Typical 18 usec Tperiod: 2.59 uV --------------------------------- Nugget diameter: 0.1" (20 nH, 320 uOhm) My 9 usec Tperiod: 498.34 mV Typical 18 usec Tperiod: 953.60 mV ----------- 1" My 9 usec Tperiod: 11.41 mV Typical 18 usec Tperiod: 12.92 mV ----------- 10" My 9 usec Tperiod: 2.23 uV Typical 18 usec Tperiod: 2.59 uV --------------------------------- Nugget diameter: 0.1" (200 nH, 32 uOhm) My 18 usec Tperiod: 235.30 mV Typical 18 usec Tperiod: 953.60 mV ----------- 1" My 18 usec Tperiod: 12.50 mV Typical 18 usec Tperiod: 12.92 mV ----------- 10" My 18 usec Tperiod: 2.59 uV Typical 18 usec Tperiod: 2.59 uV
Equal power was used for all the above examples. Also I considered the fact that 4 times as many gathered samples per second yields twice the SNR. So the above results were resulting signal from nugget multiplied times square-root of sample rate. Also, if the pulse was half the time, then I made sure the coils voltage was doubled so as to maintain the same resulting peak current on the coil. The total energy put into the coil is equal to I^2. Also if the sample rate was 4 times greater then I cut the coils voltage by half. This is because four times as many pulses occur per second. Cutting the coils voltage in half equals to one-fourth the energy thereby making it equal in power dissipation.
Here are some reasons to use my square symmetrical pulse method:
1. To work with one voltage level; e.g. 6 volts. Rather than say 12 volts for charge and 400 volts during collapse time. Or in my case it would be more like 180 mV charge and 6 V during collapse time since I'm using a 1-turn coil.
2. To simply the design. I'm not an expert circuit designer and struggle with gotchas of circuitry that you pros know about. Well, they seem like gotchas to me, lol. It is very important to recycle the collapsing magnetic fields energy. So if I went with the typical design then I would have to charge the coil at say 180 mV and then recapture the collapsing field into a 6 volt source. I could use a 180 mV battery to charge the coil, lol, and then collapse into a capacitor that's charged to 6 volts and then recycle a small amount of the caps energy back into the 180 mV battery. Or I could do the reverse and use a 6 V battery and somehow efficiently maintain ~180 mV on a cap and then use that cap to charge to coil and then collapse the field directly into the 6 V battery. Either way you are going to lose energy in the energy conversion process. Perhaps an efficient method is to have a dozen 180 mV batteries. I could use one of the batteries to charge the coil and then collapse the field into all 12 batteries. Then every so often I would have to rotate the batteries. Are there any good 180 mV batteries? In my case 600 mV sound better anyway. Perhaps another method is using a tiny 600 mV battery and a large 6 V battery. The tiny 600 mV would charge the coil and the collapsing field would go into the 6 V. Then perhaps every minute or so the circuit would automatically charge a toroid and dump its energy into the 600 mV battery. This could perhaps achieve 90% efficiency. Last year I was working on something else and achieved 85% efficiencies, but efficiency was not my goal. Another option is using a transformer, but at a glance I don't care for that idea.
3. The present design seems almost as efficient as the typical method except I do not have to waist a lot of energy working with wide voltage sources and energy conversions. I can simply charge the coil with the 6 volt battery and then collapse the field back into the same 6 volt battery. According to my calculations, "my 18 usec Tperiod" yields equally efficient results all the way up to nuggets that are a few inches in diameter. A 10" nugget is over half as efficient as the typical method. Although, according to my calculations, I'll be able to detect a 10" nugget farther than I'd ever want dig. ... 10" nuggets, lol.
A few designs ago I had a simple fix for the big nugget issue. The fix was simply shorting the coil in-between coil charge and collapsing field periods for several usec's. This allowed the nuggets current to significantly collapse during this "wait period". In other words, a voltage is applied to the coil. So now there's current in the coil. Then the voltage is removed and replace with a short across the coil. This short maintains the coils current. So if the 1-turn coil is 2 uH and the transistors resistance is 0.01 ohms, then after 10 usec the coil's current drops 5%, or after 20 usec a 10% loss. That's not too bad. Note that the nuggets current will collapse during this coil shorted period because the coils di/dt is extremely low. Then the coil short is removed and the collapsing field is directed across the same 6 V battery.
Perhaps I'll put the "coil short" feature back in if the detector doesn't scan as deep as desired, but that will yield a 5% to 10% loss. Not sure if it will pay off. For now, it seems this method is better for nuggets smaller than 1", unless you use a really efficient method of working with wide voltage sources and energy conversions. You could use the dozen-battery method and rotate the batteries, but that doesn't sound like fun. Otherwise I'm wondering just how much loss there would be. Perhaps 20% to 40%? If you know of a method that's over 90% efficient then I would consider using the typical pulse method. Where there's a will there's a way.
Peace,
Paul
hi to all
I guess the proof will be in the pudding, lol.
Comment