Replies to Minipulse Plus

[CAS]

Davor - I quickly tried your GB circuit today but I found that it cause quite a loss of sensitivity. Admittedly I didn't have much time to play with timing settings (which I may do over the next few days) but do you have any ideas on what the issue may be? I will try to get the cro on it properly and see if I can see exactly whats going on and post the results.

[KingJL]

The trick may be to use digital isolators or digital optos. They arent that expensive and have fast rise and fall times.

A couple years ago, I successfully used some iso7220m's that I got as samples ftom TI. I believe that they were only availabLe in SOIC packages.

[Davor]

Davor - I quickly tried your GB circuit today but I found that it cause quite a loss of sensitivity. Admittedly I didn't have much time to play with timing settings (which I may do over the next few days) but do you have any ideas on what the issue may be? I will try to get the cro on it properly and see if I can see exactly whats going on and post the results.

I'm very glad you did. I may only look at my completely populated PCB with sighs as I can't get it to my shack and breathe some life to it.
The loss of sensitivity is expected, especially for targets with tau near or at the famous "hole". That's the whole point of a GB - to obliterate ground and detect targets, but it comes in a package with a hole. A way to treat the hole is to adjust GB's time base to accept high conductivity targets.
If everything else is done right, I expect the GB solution to be valid for a wide range of time base settings, and that would allow for fast toggling between time bases and consequently some level of discrimination. In my GB circuit the adjustment of GB is achieved by variation of initial delay, and because it is set as a ratio of time base, it should compensate the same log-log slope of ground response delay at every setting of time base.

[6666]

Thank you.

[CAS]

I have been testing the MPP with a fast mono coil. Would there be any advantage to using a DD coil? If so what sort of inductance readings should I have for the transmit and receive coils?

[Davor]

I was playing with a PI Tx circuit with a diode, and I think I discovered a sort of "final frontier" of technology.

I was bothered by leakage currents that spoil weak target responses due to a gradient of discharging MOSFET via leakage mechanisms. After a few trials, I found that a simple network should do.
To grasp what the heck it is all about, just imagine a reverse polarised diode as a leaking capacitor, and it is connected to a larger capacitor that is being discharged through a resistor. At first it may seem a bit exaggerated, but problem is that this "signal" is superimposed to the targets responses, and because it is sloped, it spoils detection of weak signals.

There is no such thing as an ideal diode, so the only way of removing this effect is to supply a diode with a constant voltage bias right after flyback. That results with a small bias that is similar to EF, and therefore removed in Rx.

In case of a MOSFET without a diode, just to get it into critical damping makes a coil tau slope more level, and it results in longer delays for any meaningful sampling of weak signals. It can't be solved with garden variety MOSFETs, and the large capacitance of "normal" ones renders faster coils ridiculous.

A diode in junction with MOSFET's capacitances and leakage spoils weak signals directly, so there is little improvement and only for short samples.

Here goes my solution of a diode leakage gradient problem with 50us target at stepped coupling. Steps are separated 1 decade each, which results in 2 decade separation in response, as expected.

The response may be described as follows...
- initial stage is a flyback pulse seen as a spike
- it is followed by a steep slope which is a consequence of a coil ll damping resistor tau
- branching from a coil response are the target responses shooting downwards in parallel fashion

[WM6]

Thanks Davor.

Now we need Silver Dollar for implementation this idea in Minipulse schematic.

[Davor]

It would be nice to confirm these with some serious scope, as mine is a simple CRO, and at this time I'm not able to do much lab work anyway.
What stroke me as a serious issue is the fact that major limiting factor for sample delay is a coil response. Using a series diode reduces coil capacitance more than all other methods combined, and if the diode sensitivity reduction contribution is removed you end up with critically damped coil as the only limiting factor.

You can't start sampling prior to the forking point, and it is increasingly delayed for lower target responses. With common 300uH coil and carefully reducing all the capacitances, you can't expect to start sampling any sooner than ~15us, or otherwise you obliterate low level signals that become spoiled by the very coil response.

I think there s a way around this, but not with classic flyback drive.

[6666]

I have been testing the MPP with a fast mono coil. Would there be any advantage to using a DD coil? If so what sort of inductance readings should I have for the transmit and receive coils?

Try 300/300 uH for starters, beware of phasing.

[bernte_one]

who has finished this project already and can Show a Video of the machine in action

[KRinAZ]

I was playing with a PI Tx circuit with a diode, and I think I discovered a sort of "final frontier" of technology.
I have a thought here - that doesn't address the "diode leakage gradient problem" but I wonder if it may reduce capacitance and help early sampling some - since I am salvaging components from equipment (TV's and such) I get free - & so that I can recognize diodes of interest for the Tx circuit - I have started compiling a list of diodes listed as super/ultra fast, and at least 500V reverse voltage, and have found a wide difference in both capacitance (from 12pF to 185pF) and speed (from 25nS to 250nS) among them, also some (but not all) list total capacitance as well as "effective capacitance" on the data sheet - and I noticed on the data sheets that the MUR460 speed is 75nS but capacitance is not listed - I wonder what it is? and how it compares to the range? and if there is significant variance (and maybe the reason it's not listed)?
I can list the diodes I have discovered so far if anyone wants to see the list...

.

[Davor]

KRinAZ, you might want to change your way of quoting by adding your answer below or above the [quoting codes]
Regarding diodes, the capacitance is generally lower than naked MOSFETS provide, so it is usually a benefit. The capacitance changes with voltage, higher reverse voltage -> lower capacitance. It depends upon the barrier geometry, but generally it is not linear. It is almost impossible to obtain a true varicap diode nowadays, and they boasted square law capacitance change, but lots of it works in regular diodes as well. Keeping a constant reverse polarity across a diode keeps a capacitance at low values and leakage current at constant, which is the idea behind my solution.

Diode recovery speed is a completely different animal. It interacts with diode capacitance in switching applications, but it is different.

[Qiaozhi]

KRinAZ, you might want to change your way of quoting by adding your answer below or above the [quoting codes]

KRinAZ - this is a good suggestion by Davor, as clicking on "What's New?" shows all your replies as a single dot. Embedding your replies in the original post also makes it difficult to read.

[6666]

I can list the diodes I have discovered so far if anyone wants to see the list...

That could be a good reference to have thanks.

[gdogg]

just finnished my build of the minipulse plus rev c, easy build with the kit from silverdog, i think its better than the surf pi,1.2. however one prob im getting on both detectors is that they are very weak on pure gold, and aspecialy gold chains, as it cannot detect chains, anyone any ideas?