Need up to 10kv at up to an amp.

Things at the limits.

Re: Need up to 10kv at up to an amp.

Postby Jerry » Sat Aug 01, 2015 9:48 pm

Got a transformer, thats one of the things I went down to the bay area for. A guy had a couple e-beam power supply chassis that were partially parted out and still had the transformers, HV bridges, and a lot of other goodies I can reuse to build my power supply. The transformer is overkill, the power supply was rated to do 10kv at 1.4amps. And heavy... After I unloaded it from my pickup I found a site selling one and said it is 305lbs. So I have two of these transformers now. The second might make a good bug zapper.

I have a 3hp vfd I am going to hook up to it and see how it reacts.

ImageAirco Temescal CV-14 transformer by Jerry Biehler, on Flickr
Jerry
 
Posts: 573
Joined: Sun Jul 18, 2010 12:07 am
Location: Beaverton, OR

Re: Need up to 10kv at up to an amp.

Postby Doug Coulter » Thu Aug 13, 2015 11:28 am

I don't think they make bugs that big...you might want something a little smaller for that unless you like big balls of plasma added to the show.
Is this running 60hz (more or less with the VFD)? That should require some pretty large caps to keep ripple down, or some sort of series regulator. Varying voltage is likely going to play hob with focus, though I've seen designs with resistor chains to adapt focus to main beam voltage that tolerate wide variations.
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: Need up to 10kv at up to an amp.

Postby Jerry » Fri Aug 21, 2015 1:28 am

I hooked up a 3hp VFD to the 10kv, 1.4A transformer in the partially gutted electron beam power supply chassis after I rewired it according to the schematic. It actually worked, I did have to up the PWM carrier frequency on the VFD from 10 to 15khz, at the lower frequency it was drawing way too much current. The output is a little low, no load it should be around 12.5kvdc and it is actually outputting about 9.5kv. If I use the 208v tap I can probably get it to 10kv. It does overload the VFD after about 30 seconds which is no real surprise, this was mostly a proof of concept.

It does seem pretty stable, staying within +/-50v of 9.5kv but I have no idea what is going to happen when I put on a real load.

ImageTemescal CV-14 hooked to a VFD by Jerry Biehler, on Flickr

So, I have scrapped the idea of getting this supply running for the e-beam gun. Finding a tube has been very difficult so I picked up a Thermionics EBPS15, 15kw, tetrode electron beam power supply to do the job, it is good for up to 10kv at 1.5 amps. It uses a 4CX15000A tetrode for regulation. I got the power supply itself from a guy in Tennessee and the control box for it from BMI Surplus. Thermionics was able to supply me with a very complete manual so I should be able to get it going.

I do need to make the interconnecting cables and the filament transformer box. It contains the SCR controller for the filament transformer, the current sensor, and the filament transformer. Lucky for me the details of the operation of the unit are very complete and was able to figure out how to build it. I had a hall type current sensor from one of the old chassis that I picked up in california that will work for feed back and I can use the Airco filament transformer I picked up earlier. The controller uses the current sensor to detect emission current and then sends a 0-10v signal to the SSR module to control the power to the filament which limits emission. I picked up a Eurotheum 425A SCR controller off ebay which is rated for transformer coupled loads so I should have everything I need.

This will work out a lot better, it will have full regulation which includes arc detect which rolls back the current when it detects an arc, which happens often in the vacuum chamber during operation. Apparently the reason why is not fully understood.

The control box showed up today and the power supply was picked up by Yellow Freight today and is on it's way. New price on the power supply is about $40k, with the control box and shipping it is all coming to about $1200.

Strapped down and ready to go:

ImageThermionics EBPS15 by Jerry Biehler, on Flickr
Jerry
 
Posts: 573
Joined: Sun Jul 18, 2010 12:07 am
Location: Beaverton, OR

Re: Need up to 10kv at up to an amp.

Postby Doug Coulter » Sat Aug 29, 2015 9:20 pm

Yum, I like! I've had my share of under-ironed transformers here too. Most were meant for sub-second use, and only at full load at that (reduces the field in the core to have a load...). In other words, mostly X ray stuff - looks great till you want something that will go for a minute or more. I had one beaut that saturated with no load at 6v or so - rated for 240 under full load, but only for a few half-cycles. There's way too much of that kind of thing out there.
In your case, if I read right - you might have been on the high side of the self-resonance, and might try for lower, or do like spellman does and have an inductor across the primary to bring the self-resonance up. Most transformers with normal laminations kinda crap out around 2 khz, and ones with big turns ratios...lower yet. You may need a low pass filter before the transformer, and not assume it can handle your pwm frequency gracefully at all. If it's that high - a series L and C to gnd might help a lot.
It likely doesn't have the leakage inductance it would take to eat that fast pwm directly.

The one in my new fusor supply (as yet unused as it kinda scares me) is from a Korean war X ray, and will do CW all day long at 8kw/100kv+ - and even handle no load for a little bit. But...takes a come-along to even lift it out of the oil box, and self resonates right around 60 hz due to capacity in the secondary winding. Which in turn requires huge caps to keep the ripple down at 60hz - around 3k joules worth to make it 10% ripple at 20ma load (it'll do a lot more). That's bordering on rifle shell energy in my lab, under flammable oil, in a steel box that may not hold that much of a bang if something goes wrong. With the new remote capability (under construction) I may try it - but from a distance. That's just a buncha joules.
I sure am glad no one X rayed me with that pig. I'd bet it killed almost as many as it saved. 8kW changed to X rays? Whew. Probably controlled by a push button in the real MASH.

Yes, a regulator (even a horribly inefficient transmitting tube) makes a lot of sense here - The stored energy can be far less for one thing, and ripple filtered out by the regulator.
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: Need up to 10kv at up to an amp.

Postby Jerry » Mon Aug 31, 2015 2:06 am

I think I am just going to get an engine driven three phase generator to drive this. There is a guy up by seattle that has an old 1955 military generator that will 208/240 at 12.5kw for $550.
Jerry
 
Posts: 573
Joined: Sun Jul 18, 2010 12:07 am
Location: Beaverton, OR

Re: Need up to 10kv at up to an amp.

Postby Jerry » Mon Jan 04, 2016 3:07 am

Figured I would update the status of this project a bit

With the generator running well (http://bbs.homeshopmachinist.net/thread ... ator-Gloat) and the interconnects made for the power supply and control box I decided to power the thing up and see if it worked. I fired up the generator and flipped the breakers on the front panel. Filament voltage was good. After letting the tube warm up I satisfied the interlocks and hit the HV button. The volt meter immediately pegged out. Thats not good. I put a HV probe on the output and yep, full voltage with no regulation.

Looking through the schematics I see that the screen supply voltage is way low. It turned out two of the four of the screen bypass caps were shorted out. While checking voltages I also found that there was no grid voltage at the socket. This was from a open 10k, 30W resistor on the grid supply to the tube. So with no screen and no grid the tube was just a big diode. I ordered replacement caps and resistors from Mouser and installed them. Powered it all up and it works, fully variable from 0-10kv. While I was waiting for parts I picked up a used tube (4CX15000A) off ebay from a very helpful seller who used to work with these kinds of tubes in the past. It looks like I dont need the tube, yet, but it is nice to have a backup.

I have no idea why these caps shorted out or why the grid resistor blew. I guess I will have to wait and see.

So now that the e-beam supply works... well, at least I think it does. I have no way to build a 12+kw load for 10kv. Thats a lot of big resistors. I need to get the gun control working. The e-beam gun has 4 pockets that are rotated from the side, that's what that little stepper motor I stuck in a sealed container is for. To drive that I chose a Compumotor OEM750 indexer. It is a little self contained unit that has a battery backed up memory that can be programmed to do different things based on a few different inputs it had. It is actually pretty awful to program, there are three "trigger" inputs and three "sequence" inputs. The indexed can store 7 programs in it and the sequence inputs select the program in binary. The trigger inputs turned out to be totally useless for this project. When a trigger call pops up the program just stops there till one of the trigger inputs changes state according to the program. There are no "if" statements or anything. Really annoying.

It also has a couple programmable outputs which ended saving my but. I needed the indexer to home the crucible turret at startup and since the trigger inputs are useless I had to write one program for homing and one program each individual pocket. Since the outputs will sink as well as source I tied this into the "4" bit of the sequence inputs, these inputs normally pulled high. So when the indexer powers up the bit is pulled low and it is set to run sequence 011 (3). The homing program runs and then it sets the "4" bit high which shifts all the sequence inputs to the 1XX range. At this point the two lower bits will select the pocket position, 00, pocket 1, 01, pocket 1, 10 pocket 2, and 11, pocket 4. It actually worked. I had to set the other output as a "pocket good" output signal to tell the deposition controller that it is in position. Again I butted heads with the programming in this thing. When it runs a sequence it just runs it over and over so if you turn the output off at the beginning of the program and on at the end it just cycles the output as fast as it runs through the program. I added some delays into the program, it seems to work now.

To control everything I picked up a Sycon STC-200SQ deposition controller. I have actually picked up a few of these over the years for cheap on ebay. The SQ allows you to do sequencing which allows you to write recipes to automate the coating process. It has what is basically a simple PLC programming area in the software that allows you to program the 8 inputs and 8 relay outputs on the back. The manual happened to have a program in it to control one of their pocket drive units so I modified it to do the binary control that I use to control the indexer. The controller moves to the pocket with the material selected and does a ramp and soak and then opens the shutter over the source and monitors the deposition rate through an analog output to the e-beam power supply and final thickness by the shutter.

For a lot of materials you cant just let the e-beam sit in the middle of the crucible, you need to move it around the surface of the source material. To do this there are electromagnets on the gun that will sweep the beam across the surface much like the yoke on a CRT does. I had though about building my own driver for this, it can be pretty simple, just a couple H-bridges and something like an arduino. Then I got to to looking at the designs and they are a little more than that, they need to bias to get the beam in the center and stuff.

A MDC E-vap Programmable Sweep Controller popped up on ebay and I managed to snag it for $200. It worked, mostly. The backlight for the little LCD touchscreen was toast. It had a pretty large CCFL behind it and I could not find a replacement so I wired up a bunch of white leds in series and put them inside the old CCFL lamp tube. I bypassed the CCFL inverter to drive the led string with 12 volts directly. It actually worked out pretty well. it does have some hot spots where the leds are concentrated but I can live with it till I find another. I put a couple load resistors on it and it seems to work. The PSC sweep pattern is controlled from the Sycon controller with two more of the relay outputs.

Of course it is not just that easy, after getting the manual it turns out that this model cannot handle having the coils tied together or one end grounded like the sloan e-beam gun uses. I needed to isolate the electromagnets from ground. I managed to do this by making insulators from polyamide and PEEK plastic.

Now I need to add some more electrical feedthroughs to the chamber. I thought one 8 pin feedthrough would be enough but with needing 4 wires for the sweep instead of two and needing two more wires for the home switch I am up to 10 total. So that's what I need to figure out next.

Really bad video of the indexer and PSC being controlled by the Sycon. I took a blank rack unit and mounted the indexer and manual controls next to the controller.



Bad LCD Lamp

ImageMDC e-Vap Programmable Sweep Controller LCD by Jerry Biehler, on Flickr

LEDs installed:

ImageMDC e-Vap Programmable Sweep Controller LCD by Jerry Biehler, on Flickr

Bad Caps:

ImageBad caps by Jerry Biehler, on Flickr
Jerry
 
Posts: 573
Joined: Sun Jul 18, 2010 12:07 am
Location: Beaverton, OR

Previous

Return to High Voltage, High Power

Who is online

Users browsing this forum: No registered users and 3 guests

cron