A SEGMENTED LOW CURRENT CW MULTIPLIER
Posted: Mon May 30, 2011 10:11 am
I decided to change the way I build high voltage low current Cockcroft-Walton multipliers. Rather than build an N-stage multiplier to fit a given need I decided to build a lot of small 2-stage multiplier boards. When the need arises for a given voltage I can bolt several of the 2-stage multipliers together and drive them with a JKL BXA-12529 CCFL supply powered by a 12 volt 1 amp wall-wart supply. Because the 2-stage boards are small I made four (4) copies of the 2-stage multiplier on each of two boards. Then I used a diamond scribe and scribed separation lines on both sides of the boards as accurately as I possibly could. Then I headed for a sheet metal shop. They will usually charge me a few bucks to separate the boards on their bench shear and unlike using a saw there is no board loss eaten up by the saw. You will want to clean up the cut edges of the small boards as some glass fibers will be projecting ready to embed themselves into your fingers. Medium wet or dry sandpaper works well for this. Below is the pattern I etched on the boards (top and bottom). Since I have no workshop yet and all my chemicals have dried up I used “Express PCB” to do the actual board work:
The board on the right is oriented as if you had flipped the board on the left to the right about its vertical axis. I made one major mistake and you can see it near the left edge of the board on the right. The square pads are to be bolted together from the output of one board to the input of the next board so the boards must overlap a bit. As you can see there is one component hole that lies underneath this overlap area. This was a hole for an HV diode so I had to drill a hole to the right of the mistakenly positioned hole and its companion hole below to move the diode over. If you decide to build these boards make this correction in the pattern (or use my corrected pattern) so that no components will run into the overlap. Each of the boards in the photo is 3.8 in by 2.5 in. Each of the small boards is 1.9 in by 1.25 in.
I populated three boards with .01uF 6KV caps and 12KV x 10mA diodes and then tested them starting out with a single board then two in series and finally three. They are connected using .06 in x .26 in fine machine thread screws and nuts with smooth washers top and bottom to keep from ruining the copper cladding of the small square connection pads. Driven with a JKL BXA-12529 CCFL supply powered with the above mentioned wall wart they come out a little lower in voltage than the CW formula states but this is probably due to forward diode loss and other normal loss factors. Below is a table of voltages measured after each stage of multiplication:
2,800
5,500
7,400
10,300
12,700
15,000
I have tried no more than three in series at this point since these small boards were not meant to generate 30KV but maybe 1.5KV to 20KV. Any needs for 30, 40, 50KV would be better realized with a specially built multiplier since it would probably need to put out more than 10 mA. Additionally, I have not tried to drive these boards with a supply capable of more than a few mA. This would be useless since the diodes are rated at a forward current of only 10 mA.
Below is a photo of the three boards as they were after my testing.
Below it is the copper pattern of each of these three boards as it would appear if you were looking thru the boards to the bottom copper. This pattern is the corrected one with the rightmost diode pads moved over to clear the board overlap when two or more boards are connected. Input from the JKL BXA-12529 CCFL supply is applied to the left end of the board or linked stack of boards. Ground is applied to the lower left square pad and the positive 1KV is applied to the top left square pad. Positive output is taken from the lower right square pad and the ground pad at the voltages listed earlier in this document.
I have corrected the error in the complete board pattern in case anybody wants to replicate this adventure:
You will have to download the ExpressPCB application from their website to read the PCB file. Then you can load the attached file into ExpressPCB and print it on the transparency film you use to expose your boards. Just remember that the pattern is oriented as if you were looking thru the board to the bottom layer. Therefore, make sure you turn the transparency over so that the printed side contacts the board.
Below is a ZIP of the PCB file and a PDF file containing a PDF995 print-to-file bottom copper image.
The board on the right is oriented as if you had flipped the board on the left to the right about its vertical axis. I made one major mistake and you can see it near the left edge of the board on the right. The square pads are to be bolted together from the output of one board to the input of the next board so the boards must overlap a bit. As you can see there is one component hole that lies underneath this overlap area. This was a hole for an HV diode so I had to drill a hole to the right of the mistakenly positioned hole and its companion hole below to move the diode over. If you decide to build these boards make this correction in the pattern (or use my corrected pattern) so that no components will run into the overlap. Each of the boards in the photo is 3.8 in by 2.5 in. Each of the small boards is 1.9 in by 1.25 in.
I populated three boards with .01uF 6KV caps and 12KV x 10mA diodes and then tested them starting out with a single board then two in series and finally three. They are connected using .06 in x .26 in fine machine thread screws and nuts with smooth washers top and bottom to keep from ruining the copper cladding of the small square connection pads. Driven with a JKL BXA-12529 CCFL supply powered with the above mentioned wall wart they come out a little lower in voltage than the CW formula states but this is probably due to forward diode loss and other normal loss factors. Below is a table of voltages measured after each stage of multiplication:
2,800
5,500
7,400
10,300
12,700
15,000
I have tried no more than three in series at this point since these small boards were not meant to generate 30KV but maybe 1.5KV to 20KV. Any needs for 30, 40, 50KV would be better realized with a specially built multiplier since it would probably need to put out more than 10 mA. Additionally, I have not tried to drive these boards with a supply capable of more than a few mA. This would be useless since the diodes are rated at a forward current of only 10 mA.
Below is a photo of the three boards as they were after my testing.
Below it is the copper pattern of each of these three boards as it would appear if you were looking thru the boards to the bottom copper. This pattern is the corrected one with the rightmost diode pads moved over to clear the board overlap when two or more boards are connected. Input from the JKL BXA-12529 CCFL supply is applied to the left end of the board or linked stack of boards. Ground is applied to the lower left square pad and the positive 1KV is applied to the top left square pad. Positive output is taken from the lower right square pad and the ground pad at the voltages listed earlier in this document.
I have corrected the error in the complete board pattern in case anybody wants to replicate this adventure:
You will have to download the ExpressPCB application from their website to read the PCB file. Then you can load the attached file into ExpressPCB and print it on the transparency film you use to expose your boards. Just remember that the pattern is oriented as if you were looking thru the board to the bottom layer. Therefore, make sure you turn the transparency over so that the printed side contacts the board.
Below is a ZIP of the PCB file and a PDF file containing a PDF995 print-to-file bottom copper image.