Page 1 of 3

Re: Christmas Comes Early!

Posted: Fri Jun 09, 2017 5:38 pm
by Andrew Seltzman
The spellman power supplies do parallel nicely and will not sustain damage. I have paralleled 2 of the 40kV version on my fusor, see:
viewtopic.php?f=6&t=10294&start=100

One thing to remember is that these supplies are not stiff voltage sources, they are cockroft-walton multipliers; current only flows in one direction. If you parallel two of them and set them to different voltages, the output voltage will be the one of the supply set to greater voltage with that one supplying all the current. In practice, running them in constant current mode will definitely work in all conditions, but I have also found that by connecting the current program line of the "slave" supply to the current monitor line of the "master" supply and running both voltage program lines at the same voltage set point will cause both supplies to current share equally. You may need to adjust the slave supply to a slightly higher voltage to make this work, but in my case I did not need to do so.

The control loops of these supplies could be modified easily, but there is no need to do so.

Re: Christmas Comes Early!

Posted: Sat Jun 10, 2017 1:15 am
by ian_krase
Interesting. Though still not something I would try with my own supplies.

There's another result of them not being "stiff": they take time to change between voltage and current mode, and can output a much higher pulse current then the control is set to. This means they can electrocute you even if the current is set to a "safe" level, and it also means that, unlike, say, an NST, they aren't good for arcing because the pulses of very high current heat stuff up.

Re: Christmas Comes Early!

Posted: Sat Jun 10, 2017 4:21 am
by Andrew Seltzman
They do have filter capacitors on the output and thus can output a higher current pulse during an arc, but the energy storage is small. That being said since they can output 5mA or more continuously they can electrocute you, pulse or not. It's never good to intentionally arc switching power supplies continuously at a high rate, but this is not the normal operation in a fusor unless you are doing it intentionally. The PTV series of supply are an ideal candidate for a fusor, and I would recommend this model, you could probably get near 1e6n/s once it is modified for 350W.

Re: Christmas Comes Early!

Posted: Sat Jun 10, 2017 2:38 pm
by ian_krase
Do you still need a ballast resistance?

Re: Christmas Comes Early!

Posted: Sat Jun 10, 2017 3:20 pm
by Andrew Seltzman
I've never used one with my switching power supplies.

Re: Christmas Comes Early!

Posted: Sat Jun 10, 2017 3:30 pm
by John Futter
Ian
Unless you can guarentee that you will not have plasma breakdowns then I would use a ballast to limit dI/dt excursions in the psu
I will put up a photo of what happens to a multiplier stack if you let the supply continuously collapse the output voltage.
the first components to complain are the multiplier capacitors usually somewhere near the centre of the multiplier

Re: Christmas Comes Early!

Posted: Sat Jun 10, 2017 4:03 pm
by Andrew Seltzman
Ian,

As John said, the word is "continuous"; an occasional spark won't harm the supply, but if you are continuously arcing(like surface tracking on your ceramic grid insulator) the supply may be damaged. More advanced supplies designed for x-ray tubes, like the DXM series, have an arc fault counter that will shut down the supply after a given number of arcs in a time window. The PTV does not have this feature (it does have current limiting), so don't leave it arcing.

I tend to favor not using a ballast resistor since it complicates the HV component train by adding another series component between the supply and grid which then needs to be insulated as well resulting in a larger more complex system. Also consider cooling issues, lets say you want a 100k ballast resistor and have a 10mA system at 40-50kV. At 50kv, an arc/short will draw a peak current of 500mA from the capacitor. During normal operation, the resistor will drop 1kV at 10mA dissipating 10W. It's not a lot of power, but if in an enclosed/insulated system it may overheat. Also consider an RL snubber (parallel resistor and inductor), that way the resistor will snub the di/dt spike, but the inductor will provide very low DC resistance for normal operation.

It's not that hard to include either one, but then again my fusor runs very stably, arcing isn't a major concern for me, and I haven't had any problems.

Re: Christmas Comes Early!

Posted: Sat Jun 10, 2017 4:15 pm
by ian_krase
How would you make the inductor? Coil of HV wire?

Re: Christmas Comes Early!

Posted: Sat Jun 10, 2017 4:45 pm
by Andrew Seltzman
I've had good success wrapping copper magnet wire directly around the resistor from terminal to terminal. This was a snubber for a 100kV, 50A system though and the resistor was about 2ft long 1.5" dia, for a smaller resistor for a fusor you might need better insulation to prevent flashover between the windings.
IMG_20150901_134527848.jpg
IMG_20150901_134536374.jpg

Re: Christmas Comes Early!

Posted: Sat Jun 10, 2017 5:12 pm
by ian_krase
Cool. Makes sense, I guess. Does the assembly get potted or oil-onsulated, or just stay in air?


(And also what values would you use for a current controlled switched supply, contrasting with the faq which is for transformers​ mostly)?

Re: Christmas Comes Early!

Posted: Sat Jun 10, 2017 5:14 pm
by Andrew Seltzman
In that design, it was in an insulating tube in air. For your fusor you could go either way

Re: Christmas Comes Early!

Posted: Sat Jun 10, 2017 7:42 pm
by Garrett Young
Andrew,

The RL snubber is a good suggestion. I'm calculating something on the order of 100mH in parallel with a 100kohm (5us short, 50kV, and 10nF output capacitance) to limit the current pulse to 3A. Does this make sense to you?

Re: Christmas Comes Early!

Posted: Sat Jun 10, 2017 8:07 pm
by ian_krase
That's a very large value resistor, for a big ceramic one.

Re: Christmas Comes Early!

Posted: Sun Jun 11, 2017 1:34 am
by ian_krase
And also a pretty big inductor. Seems like this might need to be made with smaller components under oil or paraffin.

Re: Christmas Comes Early!

Posted: Sun Jun 11, 2017 1:09 pm
by Richard Hull
VCU (Virginia Commonwealth University) had a major issue with there grossly under powered, yet very expensive, Spellman supply on their fusor. They had to carefully select a ballast resistor due to the current limit tripping out the system when the plasma struck. Ultimately, they hit the correct value.

They have a 0-30kv 10ma model $$$$. They are limited to running at about 25 kv due to the gobbling up of voltage in the ballast. Thus, their fusion is minmal to marginal. Their good detection system was appropriated for some unknown reason and they were givien a virtually useless battery powered neutron detector of a type useful only in a real neutron field near a reactor. A complex scenario to be sure. They never seem to ask for assistance until they are up against it and it is late in the senior year. The effort is a continuing senior only project.
Thus, each new team must start cold. Not good. U of W has it right....Bring in all classes as pure volunteers letting the freshman be step-and-fetchits, machinists and low level techs. Each ascending class is allowed to become more involved at higher levels of operation and finally research.

Richard Hull

Re: Christmas Comes Early!

Posted: Tue Jun 13, 2017 10:29 am
by Garrett Young
My power supply arrived and it appears to be brand new. This particular X-version of the supply was private labeled for Hurletron. The only modifications required for 350W were as follows:

C39 0.022u/100V (i.e. KEMET R82EC2220DQ50J)
R50 4.7k 5% 1/4W

I didn't add a fan because of the short operational duration and low duty cycle for my application.

Re: Christmas Comes Early!

Posted: Tue Jun 13, 2017 4:18 pm
by ian_krase
Man, thanks for doing the homework. What type of capacitor is that, a film one? Would ceramic be expected to work?

Re: Christmas Comes Early!

Posted: Tue Jun 13, 2017 4:25 pm
by Garrett Young
A ceramic capacitor has the drawback that the capacitance changes with applied voltage so I would use a film capacitor (as used in the original design)

Re: Christmas Comes Early!

Posted: Wed Jun 14, 2017 6:53 am
by Dennis P Brown
Where did you get the info on how to up the rated current on your unit?

Re: Christmas Comes Early!

Posted: Wed Jun 14, 2017 9:27 am
by Garrett Young
Schematics included in this thread. viewtopic.php?f=11&t=4852

Re: Christmas Comes Early!

Posted: Wed Jun 14, 2017 5:40 pm
by Garrett Young
Actually, the V/mA feedback from the multiplier must be different between the 200W and 350W version.

Either R23 or R24 needs to be changed as well to adjust the gain of the opamp (currently a gain of 2, this would need to be reduced in order to increase output current).

Re: Christmas Comes Early!

Posted: Wed Jun 14, 2017 11:12 pm
by Garrett Young
Please ignore all my previous posts regarding modification.

Here is my tested to 504W (42kV @ 12mA, 1M++ n/sec) configuration:

1. JP2 Moved to Jumper Pins 2 and 3 on the enable header (this will enable the supply by default - Note: besides the additional danger this modification requires the input EMI filter to be spaced away from the housing several millimeters to accommodate the jumper in the 2-3 position)

2. R23 changed to 294kohm 1/4W 1% which decreases the current feedback op amp gain from 2 to 2/3 and increases the regulated current from 4mA to 12mA.

3. Removed CR11 (added for the X2947 version) which I believe provided OL (over load) protection.

4. Changed C39 to 0.022u/100V (i.e. KEMET R82EC2220DQ50J) and R50 to 4.7k 5% 1/4W (these are part of the current error integrator circuit and are the slower response values for the 350W version. Generally, additional tuning of the feedback and compensation loop is probably required since the supply is operating at a power greater than 350W. Some audible noise/instability can be heard and it passes through the transition from constant voltage to current. Ultimately, once in constant current mode it seemed stable.)

5. Changed the input fuse to 6A (was 4A).

6. Refer to Spellman's PTV model documentation for the 9-Pin control/monitor header. I used the reference voltage from pin 1 to set the current to the max (pin 2) and bias a 50kohm potentiometer to control the voltage (pin 4).

--Warning---
These changes are my own and not recommended. Power supply failure is very possible. A fan will only cool the un-potted components and this will place stress on the multiplier diodes. Proceed at your own risk.

Re: Christmas Comes Early!

Posted: Thu Jun 15, 2017 12:56 am
by ian_krase
Huh. Nice job.

Do you have a suggestion as to what to do for just the 330W mod? Go by the schematic?

Re: Christmas Comes Early!

Posted: Thu Jun 15, 2017 11:20 pm
by Garrett Young
Another note:

7. Remove several turns (3-4) from the inductor (L1) in series with the primary of the high voltage ferrite transformer (T2). This will allow higher peak current through the primary. (Careful removing too many turns will cause problems).

Previous warning still applies.

For 350W operation - my suggestion is to follow the schematic but either R23 or R24 needs to be changed as well in order to adjust the gain of the opamp (currently a gain of 2, this would need to be reduced in order to increase output current). The fact that the table at the end of the schematic doesn't indicate this is slightly concerning and either the current feedback (V/mA) is different between 200 and 350W models or I'm missing something.

Re: Christmas Comes Early!

Posted: Fri Jun 16, 2017 4:59 am
by ian_krase
I got mine today but wasn't able to get it operating. (Also, I don't yet have ballast parts or a working vacuum chamber to test it with, just a high voltage probe...)

The interior looks fairly different from that thread about the 40kv unit.

Am I correct that just leaving pin 7 of the connector floating should turn the unit on?