Quote from Matt Watts on February 1st, 2017, 01:09 AM

Positive.

Build the VIC circuit as we have posted in this thread and scope through each transistor.  Until you get to the TIP120, you will see a nice 50% duty cycle.  At that final drive is where you'll see the voltage shift that causes the TIP120 to lag at turn-off.  As you up the frequency, you'll get to a point where the output of the TIP120 is always on--100% duty cycle.  To me this behavior looks clearly by design.  It's a linear tapering of the duty cycle controlled by the frequency.  What that means is you can precisely control the duty cycle to PPM resolution by adjusting the frequency.  And as I mentioned before, my theory is the off-time should sync with the L2 and the on-time with the L1--the L2 being slightly higher frequency (shorter wire length).  I honestly think the wire length on those two chokes is far more important than anyone has considered to this point.  They control the exact phasing the cell sees.  Brettly posted some wavelength numbers in the centimeter range that is critical to water disassociation.  So we're talking phase shift frequencies in the gigahertz range.

@matt,
The wire-lengths of the chokes are equal in Stans Injector VIC, this is designed because the coils are on the same core-leg. The other VIC have their coils on different core legs placed further away from the secondary coil field where the mutual inductance is calculated differently.

So the DU (duty-cycle)  should always be 50% for T1-T2.

That TIP only maintains 50% DU if you pulse below 10kHz frequency at 50% DU. Also the hfe drops above 10kHz.

And adjusting the frequency makes no sense changing the DU, the VIC has only one resonant frequency.

~webmug

Posted for reference.  Not yet fabricated or populated.  Tested only on bread-board.
https://oshpark.com/shared_projects/5HX2hT1a



This version uses the LM338 VAC.  Still working on the 2N3055 VAC version.

Nice work Matt, real nice :thumbsup2:

Quote from Webmug on February 3rd, 2017, 02:44 AM

@matt,
The wire-lengths of the chokes are equal in Stans Injector VIC, this is designed because the coils are on the same core-leg. The other VIC have their coils on different core legs placed further away from the secondary coil field where the mutual inductance is calculated differently.

So the DU (duty-cycle)  should always be 50% for T1-T2.

I certainly cannot deny that fact.  Still trying to wrap my head around a meaningful mode of operation.

Quote from Webmug on February 3rd, 2017, 02:44 AM

That TIP only maintains 50% DU if you pulse below 10kHz frequency at 50% DU. Also the hfe drops above 10kHz.

What I have observed is if you pull-down the collector of Q8 with a pot, you can extend the frequency range beyond the point where the duty cycle hits 100%.  This now becomes a tunable parameter.

Quote from Webmug on February 3rd, 2017, 02:44 AM

And adjusting the frequency makes no sense changing the DU, the VIC has only one resonant frequency.

I'm not yet convinced.  There must be something at play dictating the phase alignment of the two signals going to the cell.  If this phase angle was always 0 or 180 referenced to some neutral point, I just can't see how this system could possibly work.

Quote from Lynx on February 3rd, 2017, 07:46 AM

Nice work Matt, real nice :thumbsup2:

Thank you Lynx.

I wanted to get this out there so it isn't lost.  Did way too much work on it to have it disappear.  Still more to go with the version Russ is interested in, but I think this is a good template to work from for the moment.

Quote from Matt Watts on January 28th, 2017, 11:45 PM

My suspicion is this tapered duty cycle relates to the L1 & L2 coils:  pulse-off-time effecting the L2 & pulse-on-time effecting the L1.  Just a hunch at this point.  What is clear to me though is Stan's circuit manipulates the duty cycle for a reason, else he would have used a much more simple drive mechanism.

Ronnie did mention somewhere among the hurricane of posts that impedance match has two criteria:  One when the pulse is on and the other when the pulse is off.  Again, there must be a reason and I suspect the chokes and duty cycle play into this.



I want to apologize for slacking off as of late.  My desk is in total disarray with bobbins, wire, bread-boards, parts and test equipment.  If that weren't enough, dentists and psychotic computers have zapped a good portion of my time.  Hoping to get focused again here soon.  I have a bunch of supplies to send to Russ and he has graciously offered to wind my bobbins on his Pro Winder machine.  When that is done and I have real stuff to begin testing with again, we may just get a few answers to some of my theories.
Then it's on to doing a final VIC Driver Board design and getting some boards fab'd and populated.  Still need a decent cell to test with, but one step at a time they say...

Matt, no apologies excepted.
 
we are all human here. we have things to do, it takes time... i'm in the same boat. new PC in the home lab, laptop re imagined, and the wife's PC. im  there now. everything is backed up and back on line.

must finish that EPG winder and get it shipped out. its paying some bills...

will get there. I got some extra goodies for ya... will take me more time to get them back to you.. But its good! ~Russ

Quote from Matt Watts on February 3rd, 2017, 08:05 AM

Thank you Lynx.

I wanted to get this out there so it isn't lost.  Did way too much work on it to have it disappear.  Still more to go with the version Russ is interested in, but I think this is a good template to work from for the moment.

you mean the Stan Version!

:)

You could state it that way, but since Stan is gone and none of us have access to his original equipment, I don't think there will be any Stan versions floating around amongst us.  I'll happily settle for a Russ Gries version as long as it works.   :-)

Quote from Matt Watts on February 3rd, 2017, 01:59 PM

You could state it that way, but since Stan is gone and none of us have access to his original equipment, I don't think there will be any Stan versions floating around amongst us.  I'll happily settle for a Russ Gries version as long as it works.   :-)

well... its not my version, its Stan...  Just Simplified.

END.

:)

~Russ

PS. Speak soon. another weekend... get to work! :) ~Russ

Quote from ~Russ on February 3rd, 2017, 01:52 PM

you mean the Stan Version!

:)

I allow revisions to Stan's original circuit. I support Matt's revision or "fork."

Fare enugh haxar. ;)  ~Russ

That's because of our greater understanding. :cheerleader:

I put a 2n3055 in place of the TIP120 today...Noticed a change in duty cycle with frequency, interesting

You mean on the driver side, not the VAC correct?

Yes,  put in in place of the TIP120 that drives the primary coil.

do you have a multisim file for this circuit compatible with 12?

Quote from newguy on February 5th, 2017, 04:15 PM

do you have a multisim file for this circuit compatible with 12?

Newguy,Not sure if your talking to me or Matt?

The last time I designed a drive circuit I designed it in multisim then made the PCB then found problems.
This time around it's all on the breadboard until I get a few things corrected. Until then the only thing I have was the schematic I posted earlier which was revision1. Currently working on a new revision.


Something I'm trying to understand is how Stan's drive circuit causes the change in duty cycle?
-Can anyone explain that?

HMS
Change in duty cycle  with frequency is caused by slow time of turn-on and slow turn-off time of the 2N3055 , and TIP120.
Check datasheet of these transistors.

Thanks Andy, I was not looking at the voltage amplitude controller before.

Remember, bipolar transistors "trigger" on current, not voltage like a MOSFET does.  So if you increase the drive current coming out of the 2N3906 by adding a pull-down resistor to ground, you also effectively increase the turn-off current, making the TIP120 respond quicker.  Without that pull-down, the TIP120 kind of floats during turn-off, making it lag.  This lag is what causes the duty cycle to head towards 100%.

This is not a factor of the Voltage Amplitude Control portion of the circuit.  You can set the high-side to the 12 volt rail and still see the same effect.

It's all about transistor biasing.  You'll notice on all the other transistors (Q6, Q7 & Q8), they all have biasing resistors in front of their base--little voltage divider networks.  When you add a similar voltage divider network in front of the TIP120, it's frequency response, or usable range, goes up quite a bit.  I do caution trying to get back to a solid 50% duty cycle though, as I still tend to think there as a reason for this roll-off behavior.

Thanks Matt, good explanation. I think I've figured a few ways to make my circuit do the same thing. I'll be making some changes later this week.

So, anyone been testing lately?

not yet

Quote from HMS-776 on February 6th, 2017, 08:35 AM

So, anyone been testing lately?

getting ramped up to do so as i said in my last update post...
I'm making Matt's coils... as well as mine... (again) with different wire,  so me an Matt will be able check apples to apples.
just need to get Matt a proper cell and he will be "set"
~Russ

Quote from ~Russ on February 6th, 2017, 11:22 AM

getting ramped up to do so as i said in my last update post...
I'm making Matt's coils... as well as mine... (again) with different wire,  so me an Matt will be able check apples to apples.
just need to get Matt a proper cell and he will be "set"
~Russ

Are those the two VIC cores setup or for the flat one?


I bet there are more in need of a proper cell :)  But its good to read your going to check apples to apples! :)


~webmug

there will be 4 types of core set ups.

more once we get there. its not ready yet...
Matt dose not even know just yet what ill be sending him he he...

:)

we will get to the bottom of this some tome... 

~Russ

PS. with the help of amazing fends we were able to get Matt a good set of original cores we ordered. as well as i have 2 different perm flat cores coming my way for testing ( borrowing them)....

~Russ