jpsmith123 wrote ...

I didn't think to look for a patent. Thanks for the link.

In taking a quick look at it though, it seems to me that some of the schematic diagrams are wrong. For example I don't see how the arrangement shown in Figs. 1B and 2 can work.

I thought I would see something like this:

Proud Mary wrote ...

Of course, these are just my conjectures, but I hope to have showed that once an engineering concept becomes a product, all sorts of non engineering factors have to be taken into account.

jpsmith123 wrote ...

Well here's the circuit example I posted, re-drawn in "Circuitmaker2000" (with some values put in and some resistors added):





And here's the simulated output (bottom terminal grounded):





And (if I didn't make any mistakes) here's the circuit (Fig 1B) from the Glassman patent re-drawn in "CircuitMaker2000":





And here's the simulated output from the circuit of Fig. 1B:





So, according to spice, with the same value capacitors (except for the "input" capacitor which I arbitrarily set to 5000pF), at the same drive frequency, into the same load, with the same input voltage, for a 25% increase in parts count, the circuit of Fig. 1B gives you half the output voltage, with at least an order of magnitude higher ripple, than the floating circuit.

Maybe I'm missing something, but I don't see how a grounded HV secondary coil is worth that relatively poor performance.

Anders M. wrote ...

I guess the middle-fed circuit is the most practical solution if they already have transformers made to operate with one end grounded, using bigger multiplier caps to cut down on the ripple might be cheaper than designing a new HV transformer with a floating secondary, especially for small production runs.

Anders M. wrote ...

This result really puzzled me, but I think I have it figured out now. The first circuit acts as two individual multiplier stacks, the Spellmann patent one acts as one long multiplier. The reason this makes such a big difference is that the output impedance of a multiplier is propotional to something like n^3 where n is the number of stages.

Since it acts as one long multiplier, the output voltage takes much longer to rise too, so you didn't catch the final output voltage in your 1ms simulation. I just did the simulation, and it seems that the final output voltage is around 55kV after 6ms with the same component values you used. The ripple is still very high compared to the other scenario though, so that's a clear disadvantage.

By the way, C5 in your second schematic is wrong, it's 1uF and all the other multiplier caps are 2.2nF. When I fixed this in my simulation, the final output voltage was a bit higher.

I guess the middle-fed circuit is the most practical solution if they already have transformers made to operate with one end grounded, using bigger multiplier caps to cut down on the ripple might be cheaper than designing a new HV transformer with a floating secondary, especially for small production runs.