Next gen accurate grid

For Farnsworth type designs.

Next gen accurate grid

Postby Doug Coulter » Mon Nov 18, 2013 6:37 pm

Well, I stupidly broke the last one trying to pull the HV feed through out with it still on there, duh. So, time to make another - the old one had one tungsten rod broken inside a hole, and even anodic etching in NaOH wouldn't eat the sucker out of there, so I had to make a new base. In this case, it should be a fairly large improvement, as I'm holding better tolerances (spent a whole day tuning my shop tools), and now, the holes in the baseplate don't go all the way through - just 5.5 diameters worth for the rods. This should eliminate a little issue I had with the rods sticking past the back slightly (even if you machine that flat, there are still sharp edges) and becoming field emitters. Really hard on things when that happens (and it can draw so much power I can't run on top - these TIG rods were designed to be good electron emitters).

So, 20 mil doped-tungsten rods, a new Ti base, made from "billet" (quotes so Jerry won't jump down my throat too hard - I used a piece of 1" Ti rod stock, which ain't, strictly speaking, a billet), and potentially, a new end ring, this time also Ti. We were seeing some hydrocarbons with the otherwise very good graphite end ring, I suppose from very hot D atoms hitting it and knocking off C atoms to combine with in a cooler spot elesewhere in the fusor.

Here's what it looks like:
100_2624.JPG
New grid in progress


The base is somewhat special. You'll notice I "concaved" the center. That's because there is a beam of hot, relatively high "pressure" ions that hit it dead center when it's in use. That's where most of the heat is generated. Ti stinks for heat conduction, so the object here is to get that heat generated as close as possible to the copper/BN feedthrough, so it can carry it off most efficiently. I got there with a center drill (1/8th inch at the tip), then a 3/8" drill part way, then a medium angle deburring tool to kind of chamfer the angles and let the beam hit inside the little hole, right where the 10-32 stud begins - so the heat is less than .050" from the sink.

When I get a little further, the W rods will be held in their .020" holes via centerpunching around the edge, like before - that worked quite well. And if I can get that Ti ring round (it's 1/8" wire bent to a circle and butt welded with my TIG welder), and drilled (gotta figure out how to hold it centered - I'm partway though building a jig for that), then the new other end will be that ring.

This time, insead of the circle table and mill, I used the tool-post grinder I made for my lathe, and a big degree wheel stuck on the chuck with magnets to get the alignment (I used one for cams from Summit Racing). The reasoning was I was breaking drills, a real bummer with what they cost and how long they take to get, with the mill, which won't spin anywhere near as fast, and Ti is picky about that when making tiny many-caliber deep holes, or it won't clear the chips. Looks like they came out plenty accurate - all of this is in the 1 mil error range, total. That despite not having grade 7 bearings (not hardly) in the tool post grinder - they are the cheap guys, but have no sensible play at all - the issue is how straight the quill/collet is, and vibration, which amounts to a couple mills when the drill isn't in contact with a hole.

The last time I made one close to this good, I discovered something I hadn't seen before - the beam out of the end was so finely focussed, I was making a screen I'd put inside my view window heat to orange hot, with something negatively charged.
Here's a video that shows the effect better than typed words can - watch the spot move away from the ion source grid on the right - closer to us than the main grid:


This is with the previous accurate grid. The older, all carbon but for the .040" rods is what we ran just before the 2013 HEAS, and it shows no null with ion source voltage variation - this seems to be due to the slight twist on the older grid, which was a mistake in manufacturing (perhaps, it actually works really well, as that video showed, but doesn't toss out this concentrated beam from the end as far as I can tell. Or it's not as well focused when it hits the viewport (about 15" from the end of the main grid - all the way across the big tank).

But, wouldn't you know it - right after I manage to get a really nice baseplate made, without breaking off any drill bits in holes (I did break one, but that was just by touching it a little too hard) - my dremel tool, mounted in an old dremel drill press I use for cutting TIG rods to length, took what is probably its final dump and burned up the shaft coupler (which I've remanufactured several times, but this time - big smoke, plastic all melted, it's probably toast for good now). It melted the coupler that was epoxied (by dremel) to the end of the motor shaft. So now I have a knurled shaft with all the dips still full of epoxy, and now also the wrong diameter to use the "hose trick" for a flexible coupler. Sigh...it's always something, isn't it.

So, I save 20 bucks or so on broken drill bits - and lose 200 to having to buy a new dremel and of course, the new ones don't fit into the old drill press, so...I have to buy one of those too. Soooooo close, but no cigar, yet.
Here's kind of what I was shooting for this time, but I got a little fancier with the recess in the center to really get the heat to happen in just the right place for things to live longer.
OldBase&specDrawing.JPG
The new base looks a lot like this, but doesn't have the rod holes all the way through, so the back is really smooth.


Once I get all the parts made, the Ti parts will be anodized in diet pepsi (happens to be just the right strength of phosphoric acid...) to 120v or so, which cuts conditioning arcing way down, and improves Q since TiO2 is a lousy conductor and doesn't attract ions as well - which is a waste for this, at the temperatures this reaches, we can't count on the Ti holding any D at all - it gets red hot at high power inputs.

Here's a detail of the older grid, post being broken - I may still rescue it with more anodic etching - tungsten is etched in an NaOH bath anodically, where Ti isn't affected noticeably by that. It's just a matter of getting the E field in the bath to go down that tiny hole...a real pain to jig for.
ActualGridBase.JPG
The old base. Only problem with that one is the holes having been drilled all the way through. If I can fix it, I can use it again, perhaps after making a washer to "hide" those rod-ends from the field.


So, if you're in the 6" OD fusor world, with a cylinder tank and a 1" OD grid, here's how to make a few million neutrons a second with 50kv and about 20 mA (tops - heat issues - it actually has higher Q at a couple mA, but also fewer neutrons/second).
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Next gen accurate grid

Postby Doug Coulter » Mon Nov 18, 2013 6:53 pm

One reason for trying these super-accurate, and for once, perfectly straight grids, is this strange "null" we get with them. With the "twisted" grid - the rods are about 5 degrees off due to a manufacturing issue (I forgot to clean the mill table off before mounting the circle table on it to drill the holes, so it was cocked to the side) doesn't do this.

But with these straight guys, if you let in just enough gas for the grid to "light off" at full supply voltage, with the ion source off, turning the ion source on, and gradually increasing the voltage on that grid from zero - at first, the main grid draws less current, goes through a complete null (zero draw) then builds up - it's not linear - it's not even monotonic - the main grid vs the "control grid" of our "plasma triode" here (and it has a metric crap-ton of power gain, so it's really a triode).

Since we've noticed and documented that most fusion happens right at the stability boundary - onsets of current draw being the stars of high Q (by a huge factor, like 500++), this raises some interesting possibilities. As in, making a fusor-oscillator that will flip on and off at whatever frequency it decides to - heck, I don't even necessarily require phase inversion here - since I have both signs of the slope vs the input grid voltage, and both are the same basic polarity for DC...

Look for "interesting results" pretty soon - we are for certain onto something big. We've seen hyper-high Q before, when running "on the edge" and with a large inductor as part of main-grid ballast. But it was hard to get, and we could not always reproduce it on demand - maybe half the time. Now we might find it knows best all by itself, if we hook things up right...all that transit time stuff takes care of itself in one phase or the other, and so on.

It's long been my contention that the nice stable runs most go for are actually "strange attractors" (math term) for the *worst* Q. I'm of the opinion I've proved that hundreds of times in real runs I've taken data on, and now I have far better data AQ than before, so it's time to up the game and really leave everyone else in the dust. Or so I hope...another few orders of magnitude, this could get interesting...

I also plan to take some data before and after coating my tank innards with a Ti film (evap most likely, but I might try sputtering if I can get the plumbing right for that) to hold D in the walls. An earlier experiment or few showed up to 2x more neutrons with D held in the walls till they heated up and lost it. In fact, you can see that in our last run too, even though most of the wall coating has been removed. So, we add cooling, as it only takes aobut 1-2 minutes at normal power to drive all the D out via heating up the tank. It appears we've accidentally made a "tandem" accelerator here. In other words, some of the D inside or near the grid picks up electron(s) to the point of becoming a negative ion, and being accelerated back at the tank walls at potentially double the input supply voltage. Only something like that would account for doubling the fusion we see with beam on target. I believe Chris Bradley told me that the math for this electron pickup has a higher cross section than mere fusion, which would give it a scientific basis of some sort. At any rate, it's real, we see it, and the fusion at the walls is happening only where those "beams" hit, as diagnosed by a small diameter hornyak detector I tried moving around the circumference. This is what Tyler C and I have both seen before, He was using a BTI bubble detector to get spatial resolution, and we thought it might be actual neutron beaming. I think now that the interpretation is incorrect, as I can find no theory that would indicate it is "beaming". It seems to just be "localized" which can look like a beam if you're close enough with your sensor, and both of us were close.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Next gen accurate grid

Postby Doug Coulter » Tue Nov 19, 2013 7:01 pm

We don't take no for an answer around here. So I rigged a jig to cut rods on the toolpost grinder on the lathe, which turns out better anyway...life hands you lemons, you make better grids.
So, here's what it looks like, along with the rod stock, and the copper pipe jig I use to hold things straight while I stake in the rods with a spring-loaded center punch. I've also made a jig for the ring end, so it won't compress out of round in that operation.
It's a piece of 1" iron bar, turned down to .780" at one end to just fit inside the ring while I whale away on the outside. Tight fit so it won't slip or be crushed.
I left this pic at full rez, so if you click on it, you can really see the details.
Ta-Da:
NextGenGrid.JPG
Here it is, before anodizing the Ti and staking in the rods. Just a test assembly to make sure I got all the pieces done correctly.


The point of making that Ti end ring is to eliminate the graphite, otherwise a great vacuum material. But when hot D hit it, it sputtered and made hydrocarbons, according to the mass spec. This might be in part responsible for the drop off of neutron output late in my "batch mode gas control" runs - but only part. The other benefit I hope I get is that I can "stake" the rods into this one, which I couldn't do with the graphite - I had to "tie" the rods together under tension at that end with a circumferential piece of Ta wire (what I had that would take the conditions easily - nichrome might have worked as well).
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Next gen accurate grid

Postby Doug Coulter » Thu Nov 21, 2013 7:20 pm

Moving right along, she's done (I hope). Ain't she a beaut?
NewGridFront.JPG
View from opposite the HV feed end.

This is the view from the non-hv feed end. The titanium was anodized in trisodium phosphate solution (using an Al foil cathode) at 38v, as I liked that color. More on that later.
Here's what the feed through sees:
NewGridBack.JPG
From the feedthrough end

And here are some of the tools and jigs I used:
GridToolsJigs.JPG
Stuff I used to assemble this

Didn't wind up needing those super good tweezers this time, as my fingertips with the help of a 10x stereo loupe was good enough this time - everything was machined right, and fits tight. Yes, it took time to tease in all 8 rods to the second piece, once they were all in the first one. I used the two iron pieces and the half-round copper 1" pipe to hold it all straight while staking in the rods with the spring loaded center punch. Bonus is that carbon (actually, pure graphite) ring I used on the last one. This one is all either pure Ti (grade 2) or those special doped tungsten .020" OD rods, each 2.2" long or close, for an active length of just over 2". The pipe half was laid between the jaws of a vise for staking the rods at either end. This prevented the ring end from being crushed out of round, for starters, and held it all straight once I got it all straight.

Why Ti for the ends? Well, because Ti ;)!
It doesn't outgas badly. Once it's absorbed either oxygen or nitrogen on the surface, it won't store hydrogen either - no transfer in either direction. This is a feature, not a bug, it makes it easier to control tank pressure during warmup, because untreated (sputter cleaned) pure Ti will hold enough D for about 3 fusor runs, and let it all go the first time it gets hot. In this case, it should be pretty inert. Why anodize? Well, that for sure puts some oxygen in there - and hot D won't reduce TiO2 one bit. Further, it's a pretty decent insulator, which as Kohl says, helps with arcing problems and field emission in vacuum. I've found anodizing either Ti or Al is great for that - but I can't use Al here, as I plan to beat the snot out of this one, and the Al, though a far better heat conductor, would likely melt.

In operation, there's a beam centered on the axis - the part of the poisser that isn't the radial rays. This beam - whatever it is is negatively charged as we showed above - becomes very tiny, focussed, and with crazy huge energy density where it hits the endplate, where most of the heat is generated. That's why the recess there - the heat's going to happen at the bottom of that hole, which the chamfer leading in hopefully helps defocus a little bit - the beam is about the size of a human hair normally at the lower current end of the running range we've found some interesting Q events at. It could even melt the Ti, since Ti is a lousy heat conductor, which again, is why the recess - get the heat to happen where the heatsink is going to be, so there's a very short path to get it out of there.

Since it's night and I'm off grid and on batteries, I'll wait for tomorrow to bring the tank up to STP - I'll probably use dry nitrogen for that and work really fast, since no other changes are needed this time, so I'll have sun power to run the pumps for the re-outgassing, which generally takes awhile. It takes a lot less time if I use dry gas and work fast, but some things you just can't cheat on. All these parts have been out in air for a long time, so there's certainly some junk they'll carry along.

Further, we just changed over to a new tank of D, so the D won't be very pure at first, there will be some air in it. I'll have to flush that out of the regulator, so there will be a lot of pump running before we get back to real purity - it's sitting at 8.6 e-9 mbar now - that usually takes 2 days even when there hasn't been air in the tank, for example, after a fusor run at 2.5 e-2 mbar in pure D (well, 5 9's pure according to the mass spec). I usually get below e-6 mbar within a couple minutes of the turbo getting up to full speed if the tank hasn't been open long, then the long wait commences. This time I'll be deliberately leaking impure D through the system to flush out the supply side. When I think it's reasonably pure (via the optical spectrum off the ion source gridlet - it's a pretty sensitive test), I will then let it all hit base pressure before trying real runs. 2-3 of those should have us "back in the saddle" to take real data - and compare this grid - straight, and with .020" rods, to the other one - twisted, and with .040" rods, for Q and other things. Last time this type of test was done here, the straight grid with the smaller wires showed about twice the output and Q, both, but also was a bit more difficult to get stable. It seems the twist on the other one (and maybe the contamination from the carbon ends) makes it easier to run, and more stable, but not as good when it's all settled in. Last time I ran one like this new one - we also had that weird null effect in the middle of the ion source input power curve, which we never saw with the twisted one. Here's a vid I took of that null with the other good grid:


Current started at around 15 ma, then I ran the ion source up - and the current went down below 1ma, then rose as I continuted to turn up the ion source volts/current. Not quite a perfect null, but really interesting.

Just as an observation, this plasma "triode" has power gain - about 50w in controls 2kw on the main grid, and in the case of that null - depending on which side you're on - it has gain in either phase(!). Since it has gain, it's a real plasma triode.
A PNP tube, for you semi-heads. If Phillips had had these materials back when they invented the plasma triode in around 1955 (well before more recent claims of invention in 2010 - blind leading the blind in "press release science" these days) - it might have flown and be common. The original Phillips design had sputtering issues we're avoiding here via the use of the right stuff, and a bit different operating conditions. They were running lower voltage, higher current, and a filament to provide electrons to get the discharge started at the low voltage. The original application was a low voltage power output for AM car radios...just before germanium transistors took that field over.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Next gen accurate grid

Postby Doug Coulter » Fri Nov 22, 2013 3:50 pm

OK, we got 'er done, and now it's in the tank. The new grid is a good bit longer than the old one - probably a mistake, but time will tell after breakin and purity is reached again.
I used a spacer behind the feedthrough to pull the new grid back to about the same extension as the old one was - more or less, and here it's fairly critical, with the whole "good range" in about 1/10 inch - so I might not have hit the sweet spot - I'll work on that one once things are more stable.
Here's the spacer - it's the piece of rmax between the HDPE clamp and the tank fitting:
GridSpacer.JPG
Spacer, a quick kludge.


And here's what it looks like in the tank. Doggone auto focus gets all excited about that SS screen right up front, going to have to learn how to turn that off on this camera...
GridInTank.JPG
Mounted up and ready to ride!


I then did a break in run, and some analysis, which will be a new thread under "run data" pretty soon, right now the movies are uploading to youtube.

The main change I noticed in the run was it will run at lower pressure "on the natch" eg, without an ion source. This one also throws a tight beam out of the end that heats my RF screen to incandescence, like that other accurate grid I broke. All the other phenomena are about the same, but experience has taught me to withold judgement till the thing is "broken in" and the tank is pure again, so I will. For example, the old grid, when first put in, was hard to get to 1m n/s with. Now we've made 7 million n/s with it...and can hit 4 every time we power it on.

This one hit 1.4 m n/s on the first go. Not quite the 2x I was hoping for, maybe. We'll just have to wait and see.

I DID NOT get the null I had on the previous accurate one, but for that, the ion source grid was in a different place too, and it was a different grid. So....maybe I'll try that placement/grid again.

Here's a screen shot of the data aq on this run. Nothing special, I was just playing around with the parameters, and trying not to melt anything. Stuff did get red hot...or more, close to where the Ti might melt, but at least the temperature distribution was even.
NewGrid11-22-13.png
acreenshot of first run data, more will go on the "run data" threads later.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Next gen accurate grid

Postby Doug Coulter » Mon Mar 17, 2014 7:30 pm

Post mortem. This one's out of service now, it just wasn't that great. Staking those rod into the holes in the Ti kinda messed up the accuracy "just enough", it was too long for efficiency, and when run hard, the Ti ring got so hot I feared it would slump.
The anodizing I did on the Ti went away instantly. Back to carbon ends and pure tungsten rods...the newer one did quite nicely on its breakin run. We'll see.

One possible advantage of that "slanted" rod design (well, a machining mistake that turned out well) is that it does not form such a tight beam out of the end. These really straight ones all do, and it's hard on the SS screen protecting my glass (gets red hot) and makes the glass fluoresce so it's hard to take good pix - it's hard enough already with 3 layers of glass and a screen as is. All auto-focusing cameras get real excited about the screen and focus on it, instead of what I want, and multiple reflections off the glass mean I have to wait till night time to do anything that looks half-decent already.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA


Return to Fusors, Farnsworth type

Who is online

Users browsing this forum: No registered users and 1 guest