New HV stuff, questions

Things at the limits.

Re: New HV stuff, questions

Postby Doug Coulter » Mon Oct 25, 2010 10:35 am

Well, some parts of this are easy. Mine has the same assembly at the top of the stack as yours, pretty close. That's just a filter inductor for the second string of capacitors, with some spark gaps to keep it from arcing over if the whole stack draws an arc, and some damping resistors so things don't ring so badly. That's also what all the little series R's in the stack are all about -- current limit in an arc case, which can otherwise be a bunch of amps (thousands) and enough to blow diodes. You are showing this wired as a positive output, like mine was, but I am providing another set of series caps for the AC inputs so I can turn the thing upside down because I want negative output for my fusor from this. I already have a gonzo 125kv positive stack swiped from an old Glassman after all (with my own driver, theirs was a disaster waiting to happen), and a bunch of smaller Spellman supplies that "just work".

My transformer isn't potted of course, and at some point I may take Cliff up on that offer of higher volt coils for the big Spellman transformer I have because it's a lot better made than the Wallis one appears to be (and smaller). Either is going to be much more power than I'm going to need here, and would put the "kill" in kilowatts. In my experience, it's harder to blow drivers with smaller transformers vs the driver ratings anyway, though the full wave stack connection does help prevent reverse spikes in an arc getting into the drivers and frying them. But nothing is ever perfect in that regard. Or I may use one of my smaller homebrew transformers, I've pretty much got that "going on" now and can make pretty decent ones here.

The nature of this kind of stack is that the bottom "ground" (your point C) will try to push negative as a load is drawn from the positive output, so a lead from there, with some series R to real ground will give you a current signal to monitor or use for control. It kinda-sorta looks like they are using that other diode and stuff to look at the voltage on the first stage, and extrapolate to the full stack from there, but I'll have to do more intense analysis on it to be sure. If you really believe in the bleeder resistor values in that second filter cap string, you could get a volt signal from the bottom of that by having some series R there to ground, and it might be more accurate (but with more time delay, making a closed loop harder to do). Of course in either case, you must carefully consider what will happen when there is an arc to ground off the top of the stack, the peak currents can be staggering, and a little spark gap or mov will simply go up in smoke, along with whatever it was protecting -- this is often the hardest part to get right in a full design. You WILL have arcs at some point in the process, so making things live through that is the most important thing.

I suspect that the stuff you show on the bottom has some error in the circuit tracing you did -- I've not seen 56 ohm, 100k and 1k resistors used in parallel anywhere else in these kinds of things. IT might pay to go look that over some more??? Some of that doesn't make sense at least to me.

I didn't measure any parallel R's in my stack, but then I only used a cheap DVM that wouldn't see 100 meg values at all. I guess I should now go and check that.

Cliff might chime in, but -- hey, he's the main honcho of a competing company, so let's not abuse him too much! He's been awful good to me and by extension, us, and we owe him thanks for what he's done already. Me, I just put my money where my mouth is and actually bought some stuff from them, and I'm glad I did -- it's rock solid and way worth it unless you time has very little value. One more subsystem with "no worries, it just friggin works every time". I got an SL2KW at 50kv/40 ma - way overkill for what I'm doing, but that can be nice too. Eliminates that nagging question of what would happen "if I just had a little more" once and for all, so I can get on with working on the real limitations. Half that power level would still be a good margin for what I actually do with the fusor. I'm only pursuing this project to get higher volts, not higher power, in my lashup. More than a few hundred watts just melts things and doesn't improve results at all.

BTW, you might want to put something up on the "introduce yourself" forum so we know who you are! This is intended for everyone to get a feel for everyone else, so we know at what level to ask or answer questions. Facilitates communication if we're not talking over or under one another! When we do that, no one is getting helped, which is the main point here.
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: New HV stuff, questions

Postby chabekostc » Wed Oct 27, 2010 7:45 am

Doug-

It does not make sense to me either, these resistors in parrallel. I will have to wait till Friday when I get home, and will double check the dwg against the board. I will also post a small "I'm here" then also.

Thanks for the help so far,

Charlie
chabekostc
 
Posts: 6
Joined: Sun Oct 24, 2010 8:42 am

Re: New HV stuff, questions

Postby Doug Coulter » Wed Oct 27, 2010 5:21 pm

Looks to me like the lowest left arrow on the schiz is trying to be a current monitor output. That diode across it would be forward biased, though -- something isn't right there. This would give a negative output voltage dependent on the current drawn from the stack, it's often done that way, just looking at the current needed to charge the caps on the bottom stage, where it's highest.
If that lower left diode was the other way, and a zener...it'd be more sensible (the zener simply used to protect upstream monitoring electronics). Hard to understand the need for the 100 ohm 2w R's there -- it should be a low level voltage signal. But guys who design these things sometimes get into empirical fixes for the really evil things that can happen in arc situations.

If you work out what happens with a sudden short on the top of the stack, all those caps discharge though any series R at the bottom and....the current can be real high for a moment, enough to make real troubles. Those little 4.7 ohm series resistors in the stack aren't going to limit current much at these voltages!
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: New HV stuff, questions

Postby chabekostc » Sun Oct 31, 2010 10:18 am

Doug-

I DID find some errors, esp. the diode being backward. I took the time to do a schmatic capture, wich I find forces you to be very exacting. I have attached a photo of the board, input is from the right, and goes left from there. Ground is both bottom traces, and the the 2 100 ohm output resistors are on the left.

I think it is easier to see the layout on the capure PDF.

Any help with what to add to those 100 ohm resistor will be greatly appreciated!

Thanks,

Charlie
Attachments
photo of feedback board.jpg
HiVolt Schematic of Board 1 and Feedback Ckt.pdf
(28.29 KiB) Downloaded 353 times
chabekostc
 
Posts: 6
Joined: Sun Oct 24, 2010 8:42 am

Re: New HV stuff, questions

Postby Doug Coulter » Sun Oct 31, 2010 12:43 pm

Looks to me like that's intended to be a current sense output, and that the thing doing the sensing is not necessarily hooked to what you are showing as the ground symbols, but floating and connected between the two 100 ohm outputs there. The top one should go negative with increasing current. The float off ground is common to avoid ground loop voltages induced by high current events and other noise.

Usually, one would look at this with a differential amplifier (op amp and 4 resistors for example) that ignored common mode noise, and scaled such that the output means whatever the other circuitry needs it to mean. To get some common mode range, this would not be a single supply opamp, but one with plus and minus supplies so when both inputs are near ground it's in the middle of the working range of the sense amp. Basically, the current sensed through the resistor(s) to ground will be related to the number of multiplier stages in use and be highest at the bottom of the stack when measured this way -- which makes it easier to measure. On other words, if you get x volts per ma with one stage, adding another would give you 2x the volts per ma, another 3x and so on. Probably the best way to get a calibration is to load the stack with a known R and measure the voltage there and the stack output volts, and compute the net sensitivity in volts/output ma, then design the rest accordingly.
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: New HV stuff, questions

Postby chabekostc » Sun Oct 31, 2010 1:23 pm

Doug-

I agree with the try it and find it approach. What botthers me is that it looks like the diodes that are hooked up in reverse to the stack seem to be shorted to ground with only 150 ohms, REGARDLESS of the sense circuit. Ummm, 15 KV through 150 ohms. . . . smells like someone let the captive smoke out that is installed at the factory in most all semi devices. UNLESS, I have NO idea of what is going on in CW circuits, and this is showing only what is left over after a charge cycle or other some such esoteric trick (it's got to be a trick if I don't know it!).

Back to head scratching. . . .

Charlie
chabekostc
 
Posts: 6
Joined: Sun Oct 24, 2010 8:42 am

Re: New HV stuff, questions

Postby Doug Coulter » Sun Oct 31, 2010 2:13 pm

That's why the spark gaps are there....to protect the current sense in the event of an arc. You can't make the R's bigger or they burn up under normal loads. In a 4x stack, the current through those bottom diodes is 4x the load (at least!) current, so you can't go big R there, but do all sorts of things to live through any arcs. This is also one reason most designs are push pull. That way the currents from the caps on the "AC" side of the stack discharging cancel (hopefully) in the transformer. In a single ended design, this gets blown through the transformer backwards and fries the driver electronics too! Trust me, been there. Unless the transformer is already at saturation (which it wouldn't be in most efficient designs) this can result in some multiple of the normal power supply voltage being tossed into the driver rails -- in a 10x stack, it could be 10x the normal driver voltages....very bad. So, push pull is what we do when we have any choice about it.

In an arc, it's the entire stack volts that will appear across R's to ground there....not just one stage worth. It can really make a mess if you don't think everything through (and test to find the ones you missed). It's usually not enough joules to actually burn up a resistor by heat, but arcs can blow them apart...or arc across them.

This is one reason we do this at about the highest practical frequency -- so the caps can be smaller all over, and store fewer joules to fry things with in a failure situation (arc). It's also one of the reasons we use a ballast impedance in our fusors. Dropping a multi joule pulse into one in an internal arc usually means you're going to be getting into the tank to replace things if nothing else. Without some kind of fast current limiting, the current can build up quick to the point where metal is evaporated in there, which makes the spark a better short circuit, all the way to the point of it's an arc welder with many amps at only a few tens of volts. So limiting peak currents is key to having a fusor live. Huge peak amps will make stuff fly to pieces from the magnetic fields alone!

Those diodes are there to charge up the series caps as part of the voltage multiplication, they aren't shorted to ground by anything more than those caps, through the xfrmr impedance, and then 150 ohms. That's how all multipliers work. When the transformer polarity is opposite of the desired output, we charge a series cap though those diodes, when the polarity reverses, that cap adds to the transformer peak voltage and charges the "DC" cap to the full pk-pk transformer voltage, instead of half that (peak vs peak to peak). How much current it takes to keep those series caps charged is what we are measuring here, and it will be N times the stack output current (N being the number of stages), plus some extra for losses. Because that cap also has to see the current from all the other AC caps in the stack...so at the bottom the current is multiplied by as much as the voltage gets multiplied at the top -- no magic here, it's just a fancy rectifying impedance conversion from low volts-high current to high volts-low current.
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: New HV stuff, questions

Postby William A Washburn » Tue Nov 30, 2010 8:01 am

I've been wanting to buy a Fluke 80K-40 High Voltage Probe for my Fluke-77 but don't want to pay $170.00 for a cheap-ass voltage divider. The problem is that they won't sell the replacement resistors by them selves (at least I can't find them). I've seen several of you using regular panel meters that have been modified to run 10-100 MV and another 10-40MV. Can you help me with the divider? Even at 100,000,000 ohms and 1 mA the wattage is still 100 watts for a resistor in the chain. Must be a better way to build a high-voltage voltage divider. Can anybody help me?
Thank. Bill Washburn
User avatar
William A Washburn
 
Posts: 93
Joined: Fri Oct 15, 2010 8:12 am

Re: New HV stuff, questions

Postby Doug Coulter » Tue Nov 30, 2010 10:19 am

Many of us use a B&K probe on a normal DVM, originally meant for TV service -- grab them while you can. I note I got mine (have two) at DigiKey, but they are not stocking them any more as a standalone probe. I've used these over their ratings (for short times!).

You're right, even 100 megs is no where near enough. Try more like 10 times that (eg 100 ten meg resistors in series) in a glass tube (soda or quartz, pyrex might be more conductive than the R chain), in oil. It's a real pita to make one, but I've made a couple of them. You use a piece of coax coming out the lowvoltage end of the tube, glue it in with silicone. The next day, carefully fill it with mineral oil (use a tiny funnel so you don't get oil up on the HV inside end) and seal with more silicone. Don't use tiny 1/8 or 1/16w resistors -- you know the math, but add to that they won't stand of a lot of volts apiece -- they'll arc over the body insulation. 1kv per is a decent number unless you go to Mauser and find some bulk ceramics that are HV rated, as CliffS suggests.
Such a homemade probe won't be frequency compensated of course, but you can add a little sleeve to the HV end connected to it and get close, or a cap across the LV end (that coax) depending on how it works out -- you can test that with low voltage square waves. In general, the old ohms law and power law apply, along with how much one R can hold off without arcing, or there being corona that makes it act like a different resistor than it's marked as. Which is the main reason for the oil. Do leave a little air space, and treat the thing gently after you make it, or the oil will come out.

You can sometimes find these probes at hamfests and similar places. The outer insulation is as important as the rest because there's generally going to be some corona around when in use, which can act like a resistor to whatever else...Most HV power supply companies will sell you a nice one, but be sitting down (and take your heart meds first) when they quote a price.

This should have it's own thread, it's its own batch of problems to solve.
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: New HV stuff, questions

Postby William A Washburn » Mon Feb 28, 2011 11:53 pm

I have finally completed my first CCW stack. Under the text is a photograph of the device. I used Doug’s design (with a few modifications). This can be found under the “projects” section of the [Ye Olde(r) Coulter's Smithing Home] link on the opening page. Doug was kind enough to send me the basic pattern for the stack and this is where I started. I used a pre-built LM-317 voltage regulator board (PT-PC003 Ver. 1.0) and the standard JKL BXA-12529 CCFL power supply. I found an old laptop power supply to run as the primary source of power for the stack board. Since the voltage drop across the LM317 board I used was around 3V and the old computer power supply put out 18 volts this was sufficient to assure that I could get 12+ volts the CCFL.
I placed the two smaller board modules on female/male headers so that they could be easily removed and replaced if “fried”. If you are looking for the regulator board use the PN (PT-PC003 Ver. 1.0) and go to eBay. They have been discovered and I doubt you will get the same prices as I did four months ago ($4.00 ea, shipping free).
Did some testing today and seem to get 28KV (according to my Fluke 40KV probe) across a load resistance of 150 Megohms. Ohm’s law says this is 187mA. Can somebody tell me if this small stack is capable of such performance?
Well, at least I now have a place to start with some HV experiments until I come up with the money needed to convert one of the bedrooms to a true lab or build in the back yard. This depends on waiting for some money from a trust that is supposed to be a month or so away.
If this will become the low-end first HV project for folks on this list the obvious change would be to build the LM317 variable regulator portion onto the main stack board. It might not be a bad idea to allow a wider primary voltage range by using the larger components and a heat-sink to for the LM-317. I say this although so far I’ve had no heat problems on the mini-317 board YET.
Attachments
CCW-IMG_1510.jpg
User avatar
William A Washburn
 
Posts: 93
Joined: Fri Oct 15, 2010 8:12 am

PreviousNext

Return to High Voltage, High Power

Who is online

Users browsing this forum: No registered users and 3 guests