Experimenting with the vmh3cs transmitter

[KingJL]

I can suggest a push pull topology to drive the mosfet. Winding toroids ain't my idea of a fun Sunday.

I suggest you try that with +- 150V - 200V flyback spikes that are present at the MOSFET source connection! There are some isolated drivers out there that MAY be a solution, but any that utilize a capacitive boost circuit to provide the level shift will fail (and I have tried about five of them). There are some new generation opto-isolators (e.g. UCC23513) and some new generation integrated pulse transformer drivers (ADUM3220) that show promise. I am leaning toward experimenting with the Analog Devices (ADUM3220) units sometime in the future. But first I will have a stable pulse transformer based application as a baseline.

[KingJL]

Sure, so many parameters to be considered... But once it is gotten right, then it will be great.. because another man must source correct toroid, correct wire size and correct spacings of windings and then scratch his head when inductances and performance don't match the other guy's toroid... This is what is known as "one man's toroid is another man's hell!"

I will not disagree with any of these statements (except the "correct spacings"), but gaining an understanding of toroid fundamentals and physics takes the "hell" portion out of it. It took me quite a while to get past my reluctance to dig down into the toroid fundamentals. Finally, I said "this has got to stop"... did some research and guess what... the "hell", the mystery, and the head scratching went away! Now, if I could just get away from the bulk and large footprint...!

[dbanner]

I suggest you try that with +- 150V - 200V flyback spikes that are present at the MOSFET source connection!.

That diagram simply shows a push pull scenario, and is not meant to be a full scheme. Looking forward to your exploits on the other thread. Will need to dedicate a few years to playing catch up!! Not sure I'm that much inclined. I'm getting old and long in the tooth. Ohms Law is beginning to fade from memory, far less for Thevenin's and Norton's theorems and the like.

[dbanner]

I guess you have concluded that the performance required cannot be achieved without inductively driven isolation, but that's way beyond the scope of what I will ever pursue. You engineers have such passion, maybe for me I missed the boat.

[KingJL]

... I'm getting old and long in the tooth. Ohms Law is beginning to fade from memory...

Me too (except for the fading part)... turn 76 next week!

[Altra]

I agree with all of the comments about the transformers being a pain. But after JL, MDToday and I went through the no name Chinese toroids. We have found the right materials for the cores. These can be purchased through Digikey, Mouser and probably Farnell for under a $1 a piece. Then add 4 to 8 turns for a primary and 14 to 20 turns for the secondary. You have a gate drive transformer. Unfortunately three opamps and a few transistors isn't going to get you a ground balancing discriminating pi. Or a multi-frequency vlf or hybrid detector. Thus the next level.

So why the complexity? The Vallon, Dave Johnson and John Earle circuits all have one thing in common. They use a reservoir capacitor which holds a high voltage charge. When the Tx is triggered on, 75 to 220 volts is dumped into the coil. The current ramps up in micro seconds to almost a constant state. Johnson and Vallon use the alternating flyback energy to charge the holding capacitors. The alternating flyback is why bipolar is needed. I assume a mono polar pi with a boost supply could achieve constant current? Earle's is not clear to me, if it regenerative or switch mode boost supplies? The key here is constant current in the Tx coil, at this point ferrites become less detectable. Sampling during the Tx period with an IB coil, allows us to determine ferrous/nonferrous. On a standard pi, the current continues to flow during the TX period so it is affected by the ground. The relevant patents are listed below which go into fine detail. See below JPG cut and paste from US 8,878,515 which summarizes cc vs standard pi.


From our friends
DE 101 28 849 C1 Vallon
US 8,749,240 B1 Foster, Earle, Moreland
US 8,878,515 B1 Earle
US 9,366,778 B1 Johnson
US 10,228,481 B1 Earle

[Altra]

Me too (except for the fading part)... turn 76 next week!

Happy Birthday!

[KingJL]

Happy Birthday!

Thank you very much, but let's not rush it!

[dbanner]

As for reducing the footprints, I saw this picture. Something for the future, a special purpose surface mount toroid, tailor-made for your needs.

[eclipse]

I forgot I have the Vallon boards Took the board out and desoldered one transformer.
Thicker wire measures 80uH, 6 turns. Thinner wire measures 1.25mH, ~25 turns.
Can't measure the thickness of the wires but the thick one I would say ~0.65mm2, the other is ~0,2mm2.
Ferrite ring outside diameter is 11mm. Ferrite wall thickness is ~3mm

I'll just use the vallon transformers if I have time to build something based on your design Altra.

[Altra]

Hi eclipse,

Thanks for verifying the transformer specs. JL published the specs in the Vallon thread and I also checked one of mine. Keep in mind vmh3cs is driving the half bridge with a 3.3v logic signal. Their turns ratio is 4 to 1. Multiply 3.3v x 4 = 13.2 volts gate to source. We are using 5v logic, I have settle on 2:1 ratio(10 Vgs) for now and I believe JLKing is using 3:1(15 Vgs) ?

I was pondering if the vmh3cs tx pcb has enough of the power supply to try interfacing it with a microprocessor? It appears all of the logic lines are available. Just unsure how much of the power supply is controlled by the processor pcb?

[KingJL]

Just unsure how much of the power supply is controlled by the processor pcb?

The 22V boost regulator (HA14, resistive/capacitive network beside the HA14, TL061, and IRF7470) is totally controlled by the FPGA via the signalling of the HA14 at the large inductor end of the TX board. The boost pump is held inactive during TX and the meaningful part of the RX. It is turned on and pulsed during the last 360 usec of the RX(the 360 is approximate as I am going completely on memory). I believe this is to prevent interference with/during the TX and valid RX time ( first 155 usec ). The 22V is then regulated to 20V by the two remaining TL061's, a resitive/capacitive network(s) under the two 0.010uF, 630V capacitors, and the FR9220. GND ( 0V is TX-, 20V is TX+, with common being 10V).

Annotated TX upper B.pdf

[green]

I agree with all of the comments about the transformers being a pain. But after JL, MDToday and I went through the no name Chinese toroids. We have found the right materials for the cores. These can be purchased through Digikey, Mouser and probably Farnell for under a $1 a piece. Then add 4 to 8 turns for a primary and 14 to 20 turns for the secondary. You have a gate drive transformer. Unfortunately three opamps and a few transistors isn't going to get you a ground balancing discriminating pi. Or a multi-frequency vlf or hybrid detector. Thus the next level.

So why the complexity? The Vallon, Dave Johnson and John Earle circuits all have one thing in common. They use a reservoir capacitor which holds a high voltage charge. When the Tx is triggered on, 75 to 220 volts is dumped into the coil. The current ramps up in micro seconds to almost a constant state. Johnson and Vallon use the alternating flyback energy to charge the holding capacitors. The alternating flyback is why bipolar is needed. I assume a mono polar pi with a boost supply could achieve constant current? Earle's is not clear to me, if it regenerative or switch mode boost supplies? The key here is constant current in the Tx coil, at this point ferrites become less detectable. Sampling during the Tx period with an IB coil, allows us to determine ferrous/nonferrous. On a standard pi, the current continues to flow during the TX period so it is affected by the ground. The relevant patents are listed below which go into fine detail. See below JPG cut and paste from US 8,878,515 which summarizes cc vs standard pi.


From our friends
DE 101 28 849 C1 Vallon
US 8,749,240 B1 Foster, Earle, Moreland
US 8,878,515 B1 Earle
US 9,366,778 B1 Johnson
US 10,228,481 B1 Earle

[ATTACH]48697[/ATTACH]

I've been trying to understand why the complexity. It's being done, JL's thread for one.

The key here is constant current in the Tx coil, at this point ferrites become less detectable. Sampling during the Tx period with an IB coil, allows us to determine ferrous/nonferrous. On a standard pi, the current continues to flow during the TX period so it is affected by the ground.

I'm thinking need to look at X signal to determine ferrous non ferrous, when current is changing or very shortly after. Can't determine ferrous non ferrous after current is constant at zero or some value. The JPG constant current traces https://www.geotech1.com/forums/atta...1576345487show the penny and bottle cap to be opposite polarities, true during charge and discharge not after current is constant. Ferrites are most detectable during charge and discharge. Maybe I'm missing something or there are other reasons to to do it?

Including some scope pictures. Not the same constant current(takes longer to charge than discharge)but I think show polarities for ferrous non ferrous targets.

Tx current reference(amplifier common)is 2.5V above zero current so zero current is -2.5V.

[Altra]

[/COLOR]

I'm thinking need to look at X signal to determine ferrous non ferrous, when current is changing or very shortly after. Can't determine ferrous non ferrous after current is constant at zero or some value.

Hi Green, I am no expert. I just study the patents and do actual experiments.

Pure ferrites don't develop eddy currents, so once the current quits flowing their influence on an inductance balanced coil decrease. While conductive metals develop eddy currents and continue to affect the coil balance after the current becomes steady . Iron exhibits both a strong X and lessor R component, this affects the coil balance in the opposite direction. Sorry, I can't give a better explanation.

[Altra]

The 22V boost regulator (HA14, resistive/capacitive network beside the HA14, TL061, and IRF7470) is totally controlled by the FPGA via the signalling of the HA14 at the large inductor end of the TX board. The boost pump is held inactive during TX and the meaningful part of the RX. It is turned on and pulsed during the last 360 usec of the RX(the 360 is approximate as I am going completely on memory). I believe this is to prevent interference with/during the TX and valid RX time ( first 155 usec ). The 22V is then regulated to 20V by the two remaining TL061's, a resitive/capacitive network(s) under the two 0.010uF, 630V capacitors, and the FR9220. GND ( 0V is TX-, 20V is TX+, with common being 10V).

[ATTACH=CONFIG]48699[/ATTACH]

Thanks for the information JL. It makes sense they disable the switching regulator during sampling. Tx current is probably regulated by feedback to the FPGA.