My opinion on MOSFETs

[mickstv]

I was of the understanding that a diode between the coil and the Mosfet reduced the capacitance because the coil no longer see's the Mosfet directly.

[Davor]

I don't mean to come off surly, but come on. Sit back and learn for once.

I enjoy this forum and most of the communication going on here. I also have some experience from professional radio (transmitters etc.) that may come of some use here and there. Sitting back is what I mostly do during winters as my shack is at a somewhat remote place. Now it is the way I relax in between writing some big piece. As for my detectors, I'm focused to CW detectors as they meet my requirements better than PI, but I tend to keep my options open. My next build will most probably be a simple watertight PI for diving purposes.

I did not realise you got offended, so I'll let you be and wait for some results (as most people do). Good luck.

[GT Blocker]

Davor,

I appreciate your response and now have a better understanding of my misguided assumptions. CW dectors are an animal apart from PIs. I appologize for gittin uppity.
In the PI world, small targets mean small capacitance is a must have in the food chain.

I did get a bit snarky. Sorry about that. ___ GTB

[GT Blocker]

I was of the understanding that a diode between the coil and the Mosfet reduced the capacitance because the coil no longer see's the Mosfet directly.

Hey Mickstv,

I'm pretty sure that the capacitance of the coil, cable, and MOSFET are considered as a whole. Someone correct me if I'm wrong here. "Back in the day", I think it was Reg who added a diode (15ETH06) between the R11/coil junction (the L1 damping resistor and coil junction) and the MOSFET. A resistor (56k) runs from the diode/MOSFET junction to V-. There is deffinately a thread here somewhere about it. Anyone remember where?

Is that what you mean? I'm speaking about the HHd only here.

[mickstv]

Hey Mickstv,

I'm pretty sure that the capacitance of the coil, cable, and MOSFET are considered as a whole. Someone correct me if I'm wrong here. "Back in the day", I think it was Reg who added a diode (15ETH06) between R11 (the L1 damping resistor) and the MOSFET. A resistor (56k) fed from the junction of the diode and MOSFET to V-. There is deffinately a thread here somewhere about it. Anyone remember where?

Is that what you mean? I'm speaking about the HHd only here.

Hi GT Blocker,

As in the picture below. The diode with the black arrow.

[GT Blocker]

Hey Mick,

That's it (almost) exactly, other than the diode part No. I referenced (heck if I know if it matters). Your schematic does not show a 56k resistor from the diode/MOSFET junction to V-. I don't know how much that or the diode value matter. I will be trying the idea out on this build anyway, so we'll see...

[mickstv]

I'm not really sure about the 56k idea, but most typical detectors include the diode as shown without any other parts.

It's late here, but tomorrow when I get a chance I'll short out the diode whilst monitoring the output of the first stage amp and see what happens to the minimum delay. If I notice a change and I'll post some screen shots, in a new thread.

[GT Blocker]

I'll be interested to see what you get...lemme know.

[Tepco]

This bit longer post is my “saga” about MOSFETS, fast sampling designs and similar issues and practicalities, not directly answering all questions, but maybe useful for someone, this is my opinion, based on some experience, and subject to criticism from any side (only wait for some time, I will be offline some time from now, unable to answer, terrain time).


First, about MOSFETs, 21$ for high performance SiC device is actually cheap (I did some tests with 70$ 4500V rated devices time ago). Most important technical parameters for this purpose is D\S output capacitance and breakdown voltage. Requirement for low capacitance is obvious, saving 157pF is about enough to pay for coax capacitance. Importance of high breakdown voltage is bit more technical to explain, so this is my attempt: During TX pulse, coil current ramp up to some value (determined by pulse width and coil inductance), in this process, some amount of energy is stored in that inductance. After pulse is terminated, this energy must be dumped somehow, as soon as possible, to allow fast sampling. Stored energy at turn off will produce very high voltage, short “flyback” pulse, amplitude of it may reach or exceed MOSFET avalanche level, that is, say 10-15% above is rated voltage, then transistor will start to “zenner”, dissipating pulse energy during some very short period of time. Harmless for transistor, but produce two undesired effects. First is, coil is effectively “short circuited” during this time, preventing energy release and slowing down the system, shortest possible way to release is RLC dumped network, this additional action added by transistor is less optimal and slower. Another, less important fact is that during this short breakdown period, coil current is “reversed”, generating field opposite to that during TX time. Some people mentioned that this reversal can partly undo target excitation that occurred during TX, but effect is practically negligible, too short compared to TX, I actually measured it. To obtain minimal sampling delay, this avalanche period is good to be as short as possible, or eliminated completely, leading to another design considerations (MOSFET alone will not fix everything).


To achieve fast sampling, you need to use shortest practical TX pulse due to simple reason: twice pulse with is twice coil current, but energy stored goes up as square, so you have to get rid off 4 times more energy. On the other hand, for very “fast” targets, with small TC, any too long pulse is just waste of energy, above some width you will not be able to “excite” it any more, so nice compromise is needed. There are many different opinions on this, that actually high rate of Db\Dt excite target during flyback, but this is not the case, I measured it.


For your purpose, adjust detector to some short TX, like 25-30uS, or even better, find value that still not produce avalanche on MOSFET, flyback peak at or below rated voltage. Scope and x100 probe or improvised resistor divider is needed for this. Now, using short pulses, peak current and current consumption will fell, you can compensate by increasing frequency, go to 2.5, or even 5kHz, consumption will remain normal, but you will get some gain from PI principle itself, before any signal ever reach detector electronics. Right parameter for right target, no such thing like universal detector, this may reduce response to larger objects, but your intention is something else.


Making even slightest change, anywhere in the circuit will change parameters so adjustment of dumping resistor value is absolutely necessary for any new condition or change, you will need scope for this. Monitoring front end amplifier output, you can make adjustment, but final adjustment is very critical, do it with coil, shield (if used), cable and all in place. For very fast sampling, consider changing input amplifier from 5534 to something faster, nice chips are available now, but good old LM318 is fine too (OK, offset is just rubbish, but not intended to be precise, just fast). Misadjusted dumping resistor, if it is too high in value, will produce ringing, not noticeable in air test, but making whole thing unusable or very badly performing in water or soil, too low value will increase minimal delay.
Coils are story for itself, obviously some low capacitance design is needed. Litz wire can reduce coil's own eddy current decay, also for 5-6uS consider using somewhat smaller inductance than usual, accompanying shortest pulse width. Somewhere on forum I posted simple and cheap speaker wire flat spiral coil, 8in version of it, lower inductance, around 155uH, is well capable to 5-6uS sampling, with unshielded twin lead cable and IRF740, so with your MOSFET you can afford for coax. 10In version goes below 10uS, depending on configuration.


And so on.... Not a single parameter, but right combination of compromises, as usual. I'm afraid that you are going to overstretch HHD ability this way. I tried very fast sampling, and salt water, and mineralized ground, but not simultaneously. Chance that HHD will work at 6uS underwater with that type of bottom is slim. Very sophisticated and complex ground balancing mechanisms are in use in high end detectors, simple HHD design have no such feature, and at very fast, 5-6uS sampling even these features are ineffective. I get this on neutral bottom and fresh water easily, but I smell trouble with this conditions.


Maybe best way is to try to optimize “target to environment” ratio by some experimentation and detector parameter variation, go get it most suitable for given conditions, but not golden rule here. Good luck wit this all, best regard..


(Just response to recent posts, adding diode may or may not help much, depending on actual conditions, will be forward biased during flyback if breakdown is reached, then no use of it, don't relay on this too much, use if it can make advantage in given situation)

[Tinkerer]

Hey Mickstv,

I'm pretty sure that the capacitance of the coil, cable, and MOSFET are considered as a whole. Someone correct me if I'm wrong here. "Back in the day", I think it was Reg who added a diode (15ETH06) between the R11/coil junction (the L1 damping resistor and coil junction) and the MOSFET. A resistor (56k) runs from the diode/MOSFET junction to V-. There is deffinately a thread here somewhere about it. Anyone remember where?

Is that what you mean? I'm speaking about the HHd only here.

I posted the cirquit with the 56k resistor.
The diode on the drain of the Mosfet must be ultra fast and of the same voltage as the Mosfet, to be of any use. Then the mosfet COSS is isolated. With a slow pulse repetition rate, the leakage brings the voltage of the COSS capacitor down between the pulses.
With a fast repetition rate, the voltage stays high, near the Flyback voltage.
When the Mosfet closes, this capacitance discharges fast and can cause oscillations.
To bring the voltage down between pulses, I tried the 56k resistor that worked just fine for that specific circuit, however, with different voltage and different amount of stored energy, the value of the resistor needs to be changed accordingly.

Tinkerer

[Teleno]

Hi GT Blocker,

As in the picture below. The diode with the black arrow.

What detector does the schematic belong to? I'm interested in the whole set-up. Thank you!

The diode trick is something I came up with on my own too. Check that the breakdown voltage is at least as high as that of the MOSFET's.

Eventually two diodes instead of one can further reduce the off-capacitance because all three parasitic capacitances are in series: two diodes + one MOSFET.

Is Q2 the sampling switch? I wonder what the function of diode 1N4148 and the 1M resistor are.

[mickstv]

What detector does the schematic belong to? I'm interested in the whole set-up. Thank you!

The diode trick is something I came up with on my own too. Check that the breakdown voltage is at least as high as that of the MOSFET's.

Eventually two diodes instead of one can further reduce the off-capacitance because all three parasitic capacitances are in series: two diodes + one MOSFET.

Is Q2 the sampling switch? I wonder what the function of diode 1N4148 and the 1M resistor are.

Thats one of my circuits, Q2 and a few other components around it act as a flyback blocking circuit (this is used in place of the typical 1k resistor and two diodes method). Another
user here called Moodz designed the circuit.

I haven't used this circuit for ages due to limitations. My detector now uses a micro to control Q1 and Q2, plus samples.

[Teleno]

Thats one of my circuits, Q2 and a few other components around it act as a flyback blocking circuit (this is used in place of the typical 1k resistor and two diodes method).

Is that method is a trade secret? otherwise I'd like to see the full schematic and give it a try.

I'd say Q2 may add its own output capacitance to the coil, slowing it down. Haven't you experienced such an effect?

[mickstv]

Is that method is a trade secret? otherwise I'd like to see the full schematic and give it a try.

I'd say Q2 may add its own output capacitance to the coil, slowing it down. Haven't you experienced such an effect?

The circuit had limitations....


1. It only worked with mono coils.
2. Didn't work with detector designs revolving around +/- 5 volt opamp supplies.


Your better off using a micro to control the frontend.

[Teleno]

The circuit had limitations....


1. It only worked with mono coils.
2. Didn't work with detector designs revolving around +/- 5 volt opamp supplies.


Your better off using a micro to control the frontend.

Fine but I don't know how to use Q2 for the purpose. This technique is new to me.