Announcement

Collapse
No announcement yet.

Recycling TX power

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    Originally posted by Tepco View Post
    Sorry for schematic quality, best i managed to find print it and use scotch tape. Fisher Impulse is actually bad whit small conductive fast decaying targets (jewelry etc), somewhat unstable (this is static machine), and probably most peculiar metal finding PI contraption so far, completely different from anything else and one of it's kind .TX is not square wave, but bipolar 37.5uS pulse at 5.3333kHz, quartz controlled and fixed. After TX pulse, there is no flyback bang, energy is recovered thru D2, D3, but this process is now much slower than usual .So sample is taken after another 37.5uS and last same amount of time. And this is major disadvantage of this detector, 37.5uS delay is WAY too long, (we are waging wars to get 6-8uS in normal PI detectors). Simply, eddy currents in some targets will decay long before that. Next cycle is identical, whit opposite polarity. Rx use low noise input stage, then signal is AC amplified, synchronously demodulated to get DC, and fed to "Geiger counter" type VCO. I don't have coil data.
    Tepco,
    thanks for the description and explanation.
    Some years ago I got one of the Fisher Impulse detectors for repair. Not having a schematic, I could make no sense of the manner it worked, since it was so different from other PI machines. Yes, I remember of looking for the Flyback, to indicate a functioning TX pulse, and not finding it. Fortunately the fault was a broken wire and I got it to work again without understanding it.
    Anyway, it represents the first attempt of recycling power and is interesting in that sense.

    Recovering the energy through diodes, works for the energy, but the diode noise is a problem for the RX. Using soft switching diodes like Schottky's helped, in my initial attempts of Flyback recovery.
    I am sure that the right choice of diodes (there have been big technological advances in diodes) would help.
    the other way would be to use Mosfets. Again, the switching noise is the problem, but with the many, many different Mosfet types available, the right choice probably makes all the difference.
    How can we soften the switching of a Mosfet, without slowing the switching down?

    Tinkerer

    Comment


    • #17
      Some tinkering about energy recovery PI. And combined whit ideas already mentioned in another (high voltage PI) tread.

      Just not to lose point here, "recovery" like in F1280 means just circulating large reactive current thru the coil, without energy being lost like in standard "flyback" style PI, where it is just turned into heat at different parts. But there is many more.

      Charging the coil from some (maybe high) voltage source for some pulse time will produce nearly linear current ramp, discharging it, and recovering energy thru the diodes (or anything else) will take about same time, actually bit less, due to losses. This is not typical LRC flyback network whit dumping resistor etc. So using high enough voltage and very narrow charging pulse, current can be ramped up quite quickly, and then released quickly too, fast linear ramp instead of energy wasting flyback, but fast ramp whit high dB/dT without waiting for coil recovery for sampling. Now question is: Can this actually "speed up" the coil, not actually RLC or LC resonant tank, but distributed LC, using part of the energy already stored (non-recoverable) to reverse parasitic C, beyond sampling speed possible whit normal flyback PI? (instead doing it slower like in ill-fated 1280)

      Technically easy to do, microsecond pulses at voltage similar to flyback one, recovery and fast sampling. Doable whit cheap and commonly available components. Not to mention implications to very large coils, they cannot to be built to be "fast" in conventional PI terms, physical limitations etc.

      This, in combination whit energy efficiency and high sampling frequency (integrate more samples) looks like win-win combination.

      Unfortunately, this is just tinkering, my workbench is crammed whit completely different things i have to finish to schedule, so real life tests will wait for some time, but this will be interesting for sure.

      Correct me if i'm misfired something here (posting something truly stupid is not banned by forum rules), best regard, and any idea?

      Comment

      Working...
      X