ABANDONED

[Sean_Goddard]

The title says it all. I no longer have the interest nor the time to mess with this or anything else like it for that matter. You can continue it if you like, but I won't be taking part any longer.

[Sean_Goddard]

I'm leaning towards opening up this again if anyone is interested.

Moodz, I know you have designed a digital PI, but I was wondering if we could incorporate some of the ultra low noise stuff in here into your design? One more thing we could use instead of active blanking is US10,181,720 B1 but in that case we could NOT sell PCB's or commercialise this in ANYWAY without infringing on Dave Emery's patent so if people wanted to build one it would be a no kits available job.

[eclipse]

I think there's a way to sell this - if there's a kit for sale with option for the buyer to solder in the "patent protected parts" himself and by default the usual preamp way with the 1K resistor.
Also check this thread or just have a look at the schematic

[moodz]

I think there's a way to sell this - if there's a kit for sale with option for the buyer to solder in the "patent protected parts" himself and by default the usual preamp way with the 1K resistor.
Also check this thread or just have a look at the schematic

...not sure you can do that ... I am fairly sure it’s an offence to induce another party to infringe a patent.

just sayin.

[moodz]

I'm leaning towards opening up this again if anyone is interested.

Moodz, I know you have designed a digital PI, but I was wondering if we could incorporate some of the ultra low noise stuff in here into your design? One more thing we could use instead of active blanking is US10,181,720 B1 but in that case we could NOT sell PCB's or commercialise this in ANYWAY without infringing on Dave Emery's patent so if people wanted to build one it would be a no kits available job.

Hey Sean is that “active blanking” the diode switch patent ? I am not entirely convinced that idea is novel .... Tektronix has a similar circuit from 30 or 40 years ago .... and Radars have been using them since the get go. Not sure how bolting one onto a Pi is totally new idea .. but willing to be convinced.

The front end signal chain of my PI is analogue .. the ADC , TIMING , AUDIO and PWR sync is all generated digitally so any low noise bits can help in the preamp etc.

[eclipse]

I'm saying if few hooks are left on the board everyone can install couple of diodes and resistors if they feel like they need to.

US10181720

[KingJL]

.... and Radars have been using them since the get go...

Not for active blanking... in fact, not for blanking of any type! But I suspect that in truth, you really already know that.

[Teleno]

I'm saying if few hooks are left on the board everyone can install couple of diodes and resistors if they feel like they need to.

US10181720

I've struggled with a simulation and couldn't get the circuit of this patent to work. I get a relatively high voltage at the output during the transient, coupled by the capacitance of the high voltage diode. The output is severely attenuated. The text says the prior art (antiparallel diodes) suffers from the noise of the resistor in parallel with the diodes, but the circuit of the patent has 2 resistors instead of just one! I couldn't reproduce any improvement.

[green]

I've struggled with a simulation and couldn't get the circuit of this patent to work. I get a relatively high voltage at the output during the transient, coupled by the capacitance of the high voltage diode. The output is severely attenuated. The text says the prior art (antiparallel diodes) suffers from the noise of the resistor in parallel with the diodes, but the circuit of the patent has 2 resistors instead of just one! I couldn't reproduce any improvement.

Same with me, what I tried didn't work.

[Qiaozhi]

Just because an idea has been revealed in a patent, doesn't mean that it actually works.

[KingJL]

Just because an idea has been revealed in a patent, doesn't mean that it actually works.

AMEN!!!!

[moodz]

..that patent is a variation of the the diode TR switch ... used from DC to daylight in many many circuits ....



The coil flyback switches the diodes off. Its well known in the art.

[Sean_Goddard]

Ah, but it DOES work! Read the patent, What do you notice? No description of VBias, and that is for a reason...Here's how....

What follows is NOTHING of what Dave Emery has told me about how this circuit works, in fact he refused to tell me, so here is my take on it.

It is MY interpretation of this patent that the Vbias must be a LOT greater than the wanted signal, which in this case is still in the order of 20V or so (remember we are NOT going to need a preamp with this design but straight into the demodulator). The reason for this is that the, and this is the crucial bit, HIGH VOLTAGE ultra fast switching diode will have its PN boundary region severely "compressed" by the huge flyback pulse so you need a large voltage to overcome this. In other words you need around DOUBLE the wanted voltage to kick it back into conduction.

Now comes the part that the simulations can't handle, as soon as the PN junction becomes forward biased, the thing "bounces" back the OTHER way and overshoots massively so your Vbias is NOT a steady state DC, it is derived from the flyback pulse and decays at a rate which TRACKS (or approximates - simple RC circuit) the decay curve of the signal. The signal you THEN see is the portion of the Rx signal which is NOT the flyback pulse, but everything from your setpoint downwards. This means you could set your conduction point very low and have a fast front end amplifier which won't now saturate and thus need to recover so sub 7us sample times are a reality!

Doing away with the front end series resistor reduces the noise factor hugely so you get a much better front end S/N ratio.

If you directly feed to the modulators, you can apply gain further along at almost DC frequencies so your nV/Hz figures are now stupid low, this allows much higher gain and stability that previous art PI's and allows for SILLY depths and sensitivity to small gold.

Now Moodz on to your PI. I was a bit confused to see you had combined all the sample pulses into one point. I, personally, would have kept the 4 channels separate and multiplexed them, via a CD4053, into a single 12bit ADC (like the Eagle Spectrum does ADC804) then processed these signals relatively / comparatively like Minelab do. This would give you much more flexibility as regards the way in which the signals were handled, indeed different software tweaks would allow for real-time optimisation of the machine as a whole. you could also simply do away with the demodulators and use S/H like this http://schematics.dapj.com/2005/01/s...h-standby.html and implement the motion function in software.

Just a thought.

[Teleno]

Ah, but it DOES work! Read the patent, What do you notice? No description of VBias, and that is for a reason...Here's how....

Lack of diclosure is one of the reasons to refuse a patent application. The USPTO didn't do its job.

[waltr]

I started simulating this patent yesterday in LTSpice.

Wasn't getting good results and needed to adjust both the Subber RC values and the Diode Bias.
It is the Diode biasing that is crucial to getting this to work. As with most patents I have read they only write up the concept and never the critical details. I'm having best results when Vbias is +20 to + 30 Volts with 10ma through the diodes when conducting.

I am getting OUTPUT matching the Coil decay at less than 3usec. This should be great for low TC targets.

There is also the claim that gain is not needed but unsure why. In the 'simple' PI detector we run a gain of 500 to 1000 before the integrator. This is equivalent to only 'seeing' the last 4-8mV of the coil decay. In this patent is seems we can 'see' the last 4 Volts of the decay. Again unsure without building the circuits.

Anyway, very interesting.