I'm working on building a special PI detector that has 8 coils. The idea is that these 8 coils are laid on a plastic sheet and pulled along behind the operator. The coils are fired one at a time in succession (fired in an order such that coils next to each other are not fired one after the other). The output is processed by a microcontroller that can then show the user which coils have targets under them. The point of the project is to allow someone to "sweep" large, non-trashy areas very quickly. In non-trashy areas you typically spend a lot of time swinging your detector coil over large areas with no targets in them. This device would allow you to sweep those areas and throw down bean bags on resulting target areas. Then you (or a friend) could go back and use a discriminating detector to just check the areas of known targets.
In theory it works and I have a working prototype in the lab. I built it using various designs found on the web which have very typical front ends. Unfortunately, I'm getting fairly poor depth performance. I have ways to correct some of that in software but I'm trying to figure out if part of my problem is in picking values for pulse width, frequency, current, and other levels. Keep in mind that this machine is primarily being made as a coin shooter, not for gold nuggets. Since some of these parameters are variable in software I may be able to change the machine to work for gold nuggets at a later date. Here is what I have so far:
Pulse width = 200us - I read somewhere that this is really about the maximum you want to use so I picked this. However, I'm reading that much shorter pulse widths can be used with success (50us and even lower). Battery life with 8 coils is a major concern for me. So having as short of a pulse as possible while still giving good depth is very important to me. Any thoughts on what would be reasonable and how depth changes with respect to pulse width for things like silver coins and gold rings? Can using too long of a pulse degrade depth for certain types of targets? What are some of the typical values used by commercial units?
Current = 2A - I have yet to find any information about the typical current values used for most detectors. What I do know is that higher current means better depth in most cases. But as someone else pointed out there is the law of diminishing returns so increasing the current by quite a bit may only result in a couple of inches in increased depth. I'd like to know if I can expect reasonable depth using 2A or even less than this. Again, battery life and heat are big concerns so I'd like to keep this as low as possible without sacrificing depth. Can anyone tell me what a reasonable amount of current would be for most PI units? Is 2A in the ballpark or am I way low/high for coin shooting with a 400uF coil?
Frequency = The frequency is variable but I'm looking at 200-400Hz. I pulled these numbers out of some online literature. However, these also impact battery life. Since I'm pulling this thing along the ground I won't be "swinging" the coil back and forth. So you only get one chance to catch the item as it passed under the coil. This means that frequency should be as high as possible without killing the battery. Do these values look normal/low/high for a typical coin shooting PI machine?
Sample Delay = 50us. Again, I pulled this number out of some literature. But I've also been reading about using lower sample delays (15-30us) for certain materials. My pre-amp stage can't handle 15us as the coil flyback signal hasn't settled down enough by then and the OP amp is saturated. But I could easily do 30us. Any thoughts on which would give better depth for coin shooting? I know that for gold nuggets I would need to get down in the 15us range but I'm not sure how silver coins and rings react. How does sample delay impact depth detection? I know that the longer I wait the smaller the signal is and the lower the signal to noise ratio is going to be.
Amplifier Gain = 10,000. Once again I pulled this out of some online designs. This value allows me to sample at 50us easily but not earlier than about 30us. It also means that the circuit is quite noisy. I'm using a microcontroller with an 8bit ADC in it so with VREF equal to about 2.5V you get about 10mV resolution out of the ADC. I'm starting to think that my gain is way too high and I should shoot for a gain of about 1000 and sample earlier. This would reduce noise and have my circuit sampling at a point where the signal to noise ratio is higher. Any thoughts on gain?
Thanks for the help!
In theory it works and I have a working prototype in the lab. I built it using various designs found on the web which have very typical front ends. Unfortunately, I'm getting fairly poor depth performance. I have ways to correct some of that in software but I'm trying to figure out if part of my problem is in picking values for pulse width, frequency, current, and other levels. Keep in mind that this machine is primarily being made as a coin shooter, not for gold nuggets. Since some of these parameters are variable in software I may be able to change the machine to work for gold nuggets at a later date. Here is what I have so far:
Pulse width = 200us - I read somewhere that this is really about the maximum you want to use so I picked this. However, I'm reading that much shorter pulse widths can be used with success (50us and even lower). Battery life with 8 coils is a major concern for me. So having as short of a pulse as possible while still giving good depth is very important to me. Any thoughts on what would be reasonable and how depth changes with respect to pulse width for things like silver coins and gold rings? Can using too long of a pulse degrade depth for certain types of targets? What are some of the typical values used by commercial units?
Current = 2A - I have yet to find any information about the typical current values used for most detectors. What I do know is that higher current means better depth in most cases. But as someone else pointed out there is the law of diminishing returns so increasing the current by quite a bit may only result in a couple of inches in increased depth. I'd like to know if I can expect reasonable depth using 2A or even less than this. Again, battery life and heat are big concerns so I'd like to keep this as low as possible without sacrificing depth. Can anyone tell me what a reasonable amount of current would be for most PI units? Is 2A in the ballpark or am I way low/high for coin shooting with a 400uF coil?
Frequency = The frequency is variable but I'm looking at 200-400Hz. I pulled these numbers out of some online literature. However, these also impact battery life. Since I'm pulling this thing along the ground I won't be "swinging" the coil back and forth. So you only get one chance to catch the item as it passed under the coil. This means that frequency should be as high as possible without killing the battery. Do these values look normal/low/high for a typical coin shooting PI machine?
Sample Delay = 50us. Again, I pulled this number out of some literature. But I've also been reading about using lower sample delays (15-30us) for certain materials. My pre-amp stage can't handle 15us as the coil flyback signal hasn't settled down enough by then and the OP amp is saturated. But I could easily do 30us. Any thoughts on which would give better depth for coin shooting? I know that for gold nuggets I would need to get down in the 15us range but I'm not sure how silver coins and rings react. How does sample delay impact depth detection? I know that the longer I wait the smaller the signal is and the lower the signal to noise ratio is going to be.
Amplifier Gain = 10,000. Once again I pulled this out of some online designs. This value allows me to sample at 50us easily but not earlier than about 30us. It also means that the circuit is quite noisy. I'm using a microcontroller with an 8bit ADC in it so with VREF equal to about 2.5V you get about 10mV resolution out of the ADC. I'm starting to think that my gain is way too high and I should shoot for a gain of about 1000 and sample earlier. This would reduce noise and have my circuit sampling at a point where the signal to noise ratio is higher. Any thoughts on gain?
Thanks for the help!
So far the bottom line has been that my depth readings appear to be poor. I've tweaked a lot of the circuit and coil parameters based on Eric's info which has increased depth by as much as 100%. But I'm still thinking that the depth of my machine is way low compared to other units. However, I have nothing to compare my results to. This is my first experience with PI detectors so I'm completely in the dark as to what to expect. I've tried asking before about typical "air test" depths for PI detectors and all anyone wants to say is that it is common for PI detectors to air test poorly. It's almost like they are afraid or ashamed to post actual numbers for fear of being ridiculed. I'm currently air testing a US quarter at a maximum depth of about 6" in the center of my 10" coil. This seems really low to me. Is this on par with other DIY detectors or am I way out of the ballpark? I'm not expecting to see a quarter 24" away. And I understand that the depth could improve out in the field (it could also go down due to noise, etc). So can someone give me a warm fuzzy feeling about what I should be expecting on a coin shooting PI detector with a 10" coil? Should I see 6"? 10"? 18? More? Less? My goal is to get about 8" of depth on a US quarter nearer the sides of my coil. I don't really need any more than that because the person coming along to retrieve the target will be using a standard discriminating detector which usually can't get any deeper than that anyway. Again, the goal of this system is to reduce searching time. I am not intending to use the system for deep caches. Think of it more as a "surface sweeper".