ABANDONED

[moodz]

The whole concept of diode switches is out dated .....

I use the autoswitch mosfet in my simpler PI designs and it easily exceeds the performance of a diode switch. I can share the design below ...



The resistor R2 and C3 form a simple filter to prevent noise from the gate bias V6 power supply leaking into the amplifier. For very simple PI circuits you only need the bias.

Try to avoid opamps with internal protection diodes across the inputs. ( like NE5534 ).

Obviously select the correct voltage rated MOSFET with low capacitance. Since current is minimal ..... speed, voltage and capacitance are more critical ...

The high impedance at the source terminal of the fet pinches off the mosfet channel preventing high voltage at the drain reaching the amplifier input. The bias can be set so the mosfet is normally on ( no flyback ) but off during flyback. No control circuitry is required. If there is a chance that the body diode of the mosfet may go into conduction then you will need to add protection. However for most cases the circuit works as is ... used commercially ;-).

[Teleno]

green, your link is bad: "invaid attachment".

[KingJL]

green, your link is bad: "invaid attachment".

I think you may have meant Moodz, not green.

[moodz]

.... i can see it ok but here it is again ...

[Teleno]

.... i can see it ok but here it is again ...

[ATTACH]45215[/ATTACH]

I've replaced the diode switch with this and the signal got badly degraded.

Diode switch:



Mosfet autoswitch:





I suspect the Cdg of the mosfet for the oscillations, so to test the hypothesis I placed a cap between the diode switch and ground and it caused a similar distortion. For the time being I'll stick to the diode trick.

[green]

I tried the circuit. Looked good. A little slower. Didn't expect M2 to turn off.

[moodz]

I tried the circuit. Looked good. A little slower. Didn't expect M2 to turn off.

I get quite good results with this two step damping circuit ..... R1 does the heavy lifting and R7 does the clean up ... the advantage is that the hot noisy damping resistor kept away from the input amp signal during RX.

[green]

I tried the circuit. Looked good. A little slower. Didn't expect M2 to turn off.

Dumb statement, R5 current turns M2 off. BSP89 rated for 240V, what would a be good replacement for >450V?

[moodz]

Dumb statement, R5 current turns M2 off. BSP89 rated for 240V, what would a be good replacement for >450V?

STN1HNK60. .... Used this for the blocking MOSFET with good results. There's an 800 volt version too. K80

[Sean_Goddard]

If we just revise the circuit to generate the "bias" off the output off the opamp then we end up with a workable solution that is simpler. ( works too LOL ;-) )

[ATTACH]45194[/ATTACH]

Dammit Moodz, you are a GENIUS!! I would never have thought of that )shows how much I know)!

Teleno, I see you used a CURRENT SOURCE to bias the diodes. I missed that one too (getting old and senile) but that is a workable solution too.

Dave Emery didn't give me all of the circuit just the bit in the patent. he described it but I zoned out when he started getting into the maths (not my bag).

GREAT Progress on this guys. as regards patent infringement, I think that is only if you try to commercialise the product. Don't forget patents were originally meant to DISSEMINATE information and thus encourage exactly the kind of activity we are engaged in here which is taking things to the next stage of development (in our case in a NOT FOR PROFIT manner). Who knows, maybe Dave will join in the discussion when he gets better (just had open heart surgery)!

Moods, if you use your FET damping circuit, what happens if you combine your op amp feedback to control the FET, does that give better results?

[KingJL]

Don't forget patents were originally meant to DISSEMINATE information and thus encourage ...

Not "originally", but still (at least in the US). You might want to read the latest rationale of SCOTUS on the subject by looking into Helsinn vs Teva (decided/affirmed January 2019)...
"A patent is a monopoly – an “exclusive right to practice the invention for a period of years” - granted by the government for a fixed period of time (20 years from the filing date). Patents encourage “the creation and disclosure of new, useful, and nonobvious advances in technology and design.” The granting of patents is premised on a basic quid pro quo – in exchange for inventors or companies making their inventions available to the public, the government may grant a period of exclusivity to the inventors or companies."

Note the passage "in exchange for inventors or companies making their inventions available to the public".

[moodz]

Moods, if you use your FET damping circuit, what happens if you combine your op amp feedback to control the FET, does that give better results?

..hehe. ...now look at patent AU2013101058

It certainly does ....




....

[Sean_Goddard]

AWESOME!

Now even more convinced you are a Genius!!

Dave sent me the ACTUAL circuit he used and as the YouTube videos say "It's not what you think" beyond saying Dave has missed some fairly vital parts out of the patent which, whilst you can bodge a work-around as I did, any simulations wont work without.

ALL I will say is this; THINK about the grounding and negative supplies, and what polarity each power port needs to be. George, Carl...Over to you.

I DARE not give anymore than that away, but when I built the circuit, it is so quiet I can run very large levels of gain and get a very stable PI out of it. Using a 4" coil I can detect a pinhead sized gold nugget at around 3". Now THAT I think is impressive. Not sure how that works in "real world" or in-ground scenarios.....YET...

[Carl-NC]

If critical elements needed to make the invention work have been omitted from the patent then that can be grounds for invalidating the patent via inequitable conduct. I don't see the problem though, the circuit is very simple and easy to bias correctly. Back in my old life of designing bipolar ADCs for Analog Devices we used a very similar circuit for the sample-and-hold. And, as Paul showed, practically the same exact circuit has been used in other applications. I suspect it would be a simple matter to invalidate the patent if anyone thought it was worth the cost of doing so. But it is also simple to maneuver around the patent by changing the second diode to an NPN. This also has the advantage of isolating the two bias currents.

[Altra]

Carl any idea on how temperature would affect the diode blocking method? I'm thinking in might cause drifting offsets and affects on settling time?

Your circuit "switch.jpg" simulated well. I want to try it out on a real circuit.

One my first electronics projects back in the 70's was a Radio Shack thermometer kit. It used two forward biased diodes as the temperature sensor. Thus my question.