Persistent bug in PI circuit design.

[mickstv]

whatever

[kt315]

[kt315]

[kt315]

[Teleno]

Rather than a peeing contest I was expecting some input on real improvements to the monocoil PI design.

I will propose my own:

- Delayed coil damping. The damping resistor is connected to the coil somewhere in the middle of the transient, not from the beginning.
Advantage: the first part is a fast oscilation half-cycle (a little less), then the voltage lands softly (critically damped) during the second part. Since the eponential starts at a lower level, it decays faster.

Here's the idea simulated on LTSpice. Green: my method, Blue: traditional damping.

[mickstv]

Hi Teleno, real life sim. Similar implementation to your circuit but not exactly the same.


First PIX damping resistor only.





Second PIX my version of switched damping.

[greylourie]

Teleno and mickstv. Thanks for the discussion. Focusing on the transient that occurs at tx off is important. Sharing ways to shorten the duration, will hopefully sink in amongst members and encourage movement away from traditional kilohm R input and diodes - that does nothing but hide and mask so much target response. Far better to learn how to deal with the problem at "source".

[Tinkerer]

Rather than a peeing contest I was expecting some input on real improvements to the monocoil PI design.

I will propose my own:

- Delayed coil damping. The damping resistor is connected to the coil somewhere in the middle of the transient, not from the beginning.
Advantage: the first part is a fast oscilation half-cycle (a little less), then the voltage lands softly (critically damped) during the second part. Since the eponential starts at a lower level, it decays faster.

Here's the idea simulated on LTSpice. Green: my method, Blue: traditional damping.

It is good to look at the same time at the target response. Often a scheme to shorten the delay time also diminishes the target response, resulting in no net gain.

[Qiaozhi]

Rather than a peeing contest I was expecting some input on real improvements to the monocoil PI design.

I will propose my own:

- Delayed coil damping. The damping resistor is connected to the coil somewhere in the middle of the transient, not from the beginning.
Advantage: the first part is a fast oscilation half-cycle (a little less), then the voltage lands softly (critically damped) during the second part. Since the eponential starts at a lower level, it decays faster.

Here's the idea simulated on LTSpice. Green: my method, Blue: traditional damping.

Interesting idea.
Can you please post the LTSpice files?

[kt315]

Interesting idea.
Can you please post the LTSpice files?

thinking the way might be easely released with TVS diode 150...200V between gate and drain... some capacitance gate-sourse and/or own capacitance of gate will hold mos-fet opened in time of sample.

[green]

Yes that's old traditional approach. Simple way to limit and protect front end.
In meantime different ideas appeared too.
Just one example; PI detector with two separate coils, separate coil for TX and separate coil for RX. Now, the RX part is done in different manner.
No need for resistance and also no need for clamping diodes if RX coil is projected in a way to attenuate induced current by it's morphology (less induction). Or balanced with RX coil in a way to achieve less induction.
Now, such two-coils approach is open for more ideas, special ultra fast clamping diode still can be presented there since it will not produce delay as in traditional approach.
In the old days PI detector supposed to work with longer delays and extra 8-10uS doesn't meant a thing. Irrelevant.
Now in these days majority of diyers tend to short the delay as shorter as possible to be able to catch bits of signal under 10uS. Somebody says; "there is the gold"! So that's what provoked quite new fashion of "under 10uS" run.

Made a small o0o coil for charting short time constant targets. Two Rx coils side by side(wired inverted to induction balance) with a Tx coil wound around them. I can adjust the coils for zero out during coil on but not during coil discharge. Should I be able to get zero out during discharge? So far using a diff amp and induction balance seems to work best for me. No diodes or series resistors. Back to back diodes across the coil don't seem to work, probably because I can't get zero volts during discharge.

[6666]

Interesting test Green, MikeBG proposed something like this a while back, if you are useing a 5534 as a preamp, according to the data sheet the ne5534 actually has internal diodes across pins 2-3.

[green]

Interesting test Green, MikeBG proposed something like this a while back, if you are useing a 5534 as a preamp, according to the data sheet the ne5534 actually has internal diodes across pins 2-3.

Using diff amp in http://www.geotech1.com/forums/showt...967#post205967. Some improvements were suggested. Haven't tried them yet, still using the first one. Mike shows the center tap of Rx grounded. Tried both grounded and open. Don't remember seeing a difference, maybe I should.

[Carl-NC]

There are a lot of things being tossed into the mix that have nothing to do with the OP.

First, let's talk about the resistor-diode clamp. No, it doesn't play a big role in sampling time; a 1N4148 has a depletion C of about 5pF, so 2 diodes are 10pF; with a 1k resistor this is a 10ns time constant. Typically this is 100's of times faster than the coil itself, as well as the preamp overvoltage recovery. Even if diffusion capacitance is the culprit it's probably not the dominant slow thing. This clamp is commonly used in PI detectors that go down to 10us or less. It does have an impact on thermal noise, and can dominate that. Mick's solution gets rid of thermal noise.

Second, damping is another separate issue. Yes, the R used in the clamp reduces the damping R during clamp but you can account for it in the design. Yes, you can switch in the damping R during decay but it's really squirrely. I did this many years ago, it worked great in Spice but not so well in practice. Too much decay variation over ground.

Third, neither of these have anything to do with ground balance or target holes, or suitability for prospecting in harsh soil. Those are separate issues again. Yes, you want faster sampling for smaller nuggets, preferably less than 10us, but it's a holistic design that includes the coil, coil switch, damping, clamping, and preamping. On top of that you add ground balance, and hopefully deal with the target hole.

[mickstv]

A couple other things I noticed with the circuit I mocked up, re switched damping resistor, the damping resistor did run cooler. The cooler running I guess may give an improvement if heat causes extra noise in the frontend.

Also if you limit the max flyback voltage the benefit between the normal damping resistor and the switched damping resistor methods are minimal. Less than 100-200ns faster.