C+W HV multiplier

If you can draw the schematic without crossing 9nterconnections, and the component body sizes scale with the schematic symbols, you've got an smd layout.

For CW, even through-hole layouts are pretty straight forward.

However if your minimal real estate doesn't have the same general shape - say it's square instead of long - or there's some other kind of arbitrary restriction - you will still end up jumping through hoops.

RL

Such an astute observation. ...Jim Thompson

-- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | San Tan Valley, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

| 1962 | I love to cook with wine. Sometimes I even put it in the food.

What sort of voltage levels can be achieved using SMDs in this way?

CW stages don't see accumulated stress between stages - only the graduated body and end terminals of the assembly carry this. One of the advantages of the topology.

RL

With the MMBD5004 dual diodes that I'm using, you can make about 400 volts per section, a section being one dual diode and two caps. I'm running 4 sections, closed-loop regulated to 1400 volts and some intermediates.

You can stack many sections. The output current capability drops, because all those caps are in series. No part sees more than 400 volts.

There are higher voltage SMD diodes and caps around. Scale it up to use kilovolt diodes and caps, and you could make a good fraction of a megavolt.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

IME the effective limit was 10-15 stages or so. After that the losses seemed to be too high and the voltage stopped increasing. This was with through hole parts and I think 10kV per stage (it was 25 years ago so might be mis-remembering).

--

John Devereux

The accelerator folk (Cockcroft and Walton among them) who made these, used 50 to 60 Hz AC (and later, did some third-harmonic generation). The first one, got 600 kV or so, five stages. The big one at Fermilab (until 2012) was about 18 stages (three-phase). It was, as best I can tell, about the same output voltage.

In air, of course, those required a two-story room, and you can't miniaturize it much by using tiny SM components...

How about a really long skinny PC board? Add some LEDs in series with the HV diodes.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

W.jpg

We have a Glassman 250 KV power supply, it must have at least 50 stages, stands about 6 feet tall.

Jon

Banana jack output?

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Chuckle, tears.. George H.

I just pulled an ordinary 1206 10M resistor from stock and I'm running it at 2KV. Seems fine so far. That's 0.4 watts, so any higher voltage might kill it thermally.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Tell us about your transformer and driver. Q11 looks pretty large.

--
 Thanks, 
    - Win

Looks like you cold compress the layout somewhat, mmbd5004brm has two diode pairs in one sot-23-6.

--
 Thanks, 
    - Win

Cold compress?

At 400 volts per diode, and 35 mils between pads, that's 11 volts/mil creepage gradient. I was designing my layout for 8.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Here's the PCB, almost done:

and the power supply sheet:

I'm using eight 2SK4177 mosfets already, so I used one more in the flyback converter. We're into BOM minimization to reduce feeder loading.

I just finished highlighting every node on the schematic that has any voltage above 24, and checking the layout for appropriate surface and internal clearances. Near the center, I have a test point at +1400 volts, and not far away is another at -540. They are 440 mils apart, only 4.4 v/mil, which should be fine.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

The output voltage is fixed, not adjustable?

--
 Thanks, 
    - Win

There are two channels, pulse generators that are independently programmable from 0 to +1200 volts. Each channel is powered from the raw +1400, and has a 0-1200 linear regulator and then a pulser. The loads are PLZT electro-optical things, under 50 pF, and the rep rate is KHz, so power dissipation will be low, maybe 5 watts for the entire board.

I need pretty good agility on programming the pulse amplitude, up or down, so programming the switcher would be too slow. Hence the linear regs, which are actually amplifiers I guess.

This is a blast from the past. All linear, no FPGA, no uP, and I did the schematic entry and PCB layout myself. Lots of fun testing (blowing up) parts with high-voltage power supplies.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

The nice thing about the 2sk4177 is its small size. It's not physically small, but it has a small die, Ron = 10 ohms. This means it has low capacitance, with Coss under 30pF above 100V.

In my table of high-voltage MOSFETs, it's one of the smallest. Manufacturers seem to think everything must be bigger, higher current, more power. My table has 21 parts that are bigger.

There's a smaller 1.5kV part I like, the 2sk1412, with Ron 140 ohms and Coss 12pF at 30V. ON Semi offers the curious ndfpd1n150c with an attractive Coss 9pF.

--
 Thanks, 
    - Win