Kind friends

Once again LtSpice has me beat

Attached is a schematic file showing two series connected, independently excited, full wave 'voltage' doublers --The problem is that for non-zero values of R[bond] the results (i.e. E plots of 'Ea' and 'Ek') become increasingly erroneous with resistance until at ~ 500kΩ the simulation bombs altogether...

Any insights/workarounds will be greatly appreciated!

At your mercy
HP

 

Addendum: FWIW the intended function of R[Bond] is to establish a ground reference as per LtSpice's proclivities...

Again, many advance thanks!

 

Try adding a connection from each voltage source (-) to ground.

 

Your sources are set for 25,000 volts peak AC at 30kHz (interesting to find such a source in real life) .
What output do you expect?

 

I think it's a full wave voltage quadrupler but voltages look unusual.
I redrew the circuit.

 

Try this file. I deleted the lower portion of the circuit and replaced it with a copy of the upper part.
It performs the simulation at a crawling speed, but at least it's not reporting any errors.

 

Try adding a connection from each voltage source (-) to ground.

Thanks! Your suggestion works --- Attached is the circuit revised as per your advice...

Why does it fail to simulate correctly as regards the original schematic?

Best regards
HP

 

(interesting to find such a source in real life)

The sources are each secondary of a pair of LOPTS modified such that their EHT windings are 'floating' (i.e. well insulated from the core -- each end) an example of which is pictured below...

What output do you expect?

The actual output is a scant 100kV (Ea to Ek) with insignificant (~400V) ripple when loaded to 10 mA.

Modified LOPT

 

I think it's a full wave voltage quadrupler but voltages look unusual.
I redrew the circuit.

Actually it's two independently excited full wave doublers connected in series -- The advantage of said arrangement being much lower EMF drop with loading (i.e. vastly improved current availability/ load regulation)...

 

Try this file. I deleted the lower portion of the circuit and replaced it with a copy of the upper part.
It performs the simulation at a crawling speed, but at least it's not reporting any errors.

Thank you for your efforts! -- Unfortunately it is yet simulating incorrectly --- LtSpice Grrrrr!

 

OBTW --- In case I failed to make it clear -- the circuit as shown in post #1 works in reality (i.e. it produces a clean [scant] 50kV each side of Gnd) for slightly less than 100kV (Ea to Ek) It's merely not simulating correctly (which I am eager to understand inasmuch as, despite it's proclivities, LtSpice is a highly useful tool!

 

LtSpice is a highly useful tool!

And it's also free... that makes it my kind of tool...

 

OBTW --- The circuit simulates as per reality when the common point is grounded (i.e. R[Bond] is bypassed)

 

Try adding a connection from each voltage source (-) to ground.

Thanks! Your suggestion works --- Attached is the circuit revised as per your advice...

Why does it fail to simulate correctly as regards the original schematic?

Best regards
HP

Oops! Spoke too soon! -- Now the simulation runs wild when R[Bond] is opened (removed) --- 4.4MV! - If only....

 

Please have a look at this modification with attention to current through R1 and Rload ---- That fact that I(Rload) is clean whereas I(R1) is unrealistically noisy would seem to suggest LtSpice is incorrectly reacting to phase differences in the signal sources?

Merely grasping at straws

Best regards
HP

 

Are the voltage sources actually transformer secondaries?
and, if so, is each xfmr output meant to be 25kv Pk-Pk?

 

Are the voltage sources actually transformer secondaries?
and, if so, is each xfmr output meant to be 25kv Pk-Pk?

Correct - Said transformers are modified such that the secondaries are isolated from the cores/frames to at least 100kV -- FWIW the drivers are standard resonant Royer topology (hence sinusoidal output) -- as a practical matter 500w is about the limit, hence a 20MΩ (i.e. 5mA) load @ 100kV would be more reasonable.

As implemented, the two doublers are connected via a 0-20 mA D'arsonval indicator (represented on the attached schematic by a 2Ω resistor)... The attached version both works and 'simulates' albeit less than 'elegant'... For safety the 'ungrounded' side of the indicator is returned to ground via an MOV (not shown)...

I'm keen to understand the 'mechanics' of LtSpice's 'quirks' in this regard such that I may know what to expect in future simulations --- Any further input is both welcome and earnestly requested!

Best regards & many thanks!
HP

 

Hi

Give this one a try...

 

I'm keen to understand the 'mechanics' of LtSpice's 'quirks' in this regard such that I may know what to expect in future simulations

HP what part of digital model not perfect to analog world do you not understand? Discontinuous investigation of continuous phenom would need infinite data and time! So lighten up it not end of world! Bad sim run maybe cuz of nonlinear diode junctions or like you say phase difference of ac exciters at critical sample points! Who knows? Who cares? I say be happy psu works in real world is good enough! Also why grounded ma meter not elegant?

 

Hi

Give this one a try...

Indeed, single transformer/single driver arrangements have advantages, especially in the way of component count and maintenance of symmetry -- Unfortunately such would require fabrication of a 'custom' core - a service which, as far as I have been able to discover, is unavailable for small quantities... (If curious, please see post #8 for an image of the physical transformer topology)...

Inasmuch as the circuits function correctly (in reality), I am principally interested in understanding the nature of LtSpice's difficulty with the 'problem' schematics --- I require sufficient knowledge of LtSpice's 'limitations' such that I may confidently interpret results for complex designs where inaccuracies may not be self evident.

Sincere thanks for your time and effort on my behalf!!!

Best regards
HP