Post Go back to editing

High Voltage Power Supply (2.5 KV+)

I need some help putting together a circuit for a power supply capable of 3000 KV / 200μA.  I found the design reference for the LT1304 capable of 1000V, but VIN needs to be capable of supporting 12VDC.  I've also seen DN1047 with a 1000V supply but I've read on the forums that adding a CW voltage doubler would probably not work due to current spikes.

My current goal is to use a power IC connected to a transformer with the secondary output connected Cockcroft Walton voltage multiplier with 6-8 phases.  I'm able to successfully get things working with the LT1304, but I need to operate with a VIN of 12VDC. Is there another IC similar to the LT1304 that supports a higher input voltage?

I've tried 10-15 different LT ICs in LTSpice, but all of had huge current spikes when trying to ramp up the output voltage.  When using the current limit function (usually on the drain of the MOSFET) I can limit the current on the primary side of the inductor to safe levels, but the consequence is there is not enough energy going into the transformer due to the small minimum on times of most of the current Linear Tech ICs, ranging around 110-300μs.  

So is there a Linear Tech IC that has a larger minimum on time that would be suitable for this application?  Is there another design reference of a circuit that supports voltages in the 2.5-3 KV range?  Any help is greatly appreciated.

  • Hi,

    You will need to use a step-up transformer and some charge pump circuits to achieve such high voltage.

    Try the LT3580, or LT8361. 

    Below is an application with the lt3580. The transformer is crossed out because it is obsolete. You will have to provide your own.

    It is important the application runs in deep DCM, to avoid the leakage inductance spikes prematurely tripping the current limit.

    And yes, you will need several charge pump circuits similar to the lt1304 app. 

    If you decide to go with the lt8361, please connect your charge  pump stages with a resistor in series with the diode, as seen below.

    Also, ensure operation in deep DCM, and use a transformer with very low leakage inductance.

    The circuit below is only for illustration of the charge pump stage. The IC shown here has a switch current that is probably too  low for your needs.

    If you believe in luck, I wish you the best of luck, as 3KV would not feel good to touch.

    Best Regards,

    DV

  • Thanks for your help.  I'm running to issues with both of these regulators.  One of two things happen, either the current spikes on the primary side of the inductor are way beyond on the SW rating (10A+) or if I limit the current going into the primary side of the inductor (with a resistor), to keep the current spikes under the SW current limit, the IC goes into maximum duty cycle, which does not have a enough switching to keep current flowing from the primary side to the secondary side of the transformer.  Reading more into the electrical characteristics, both of these ICs have small minimum on time and maximum off time.  

    My theory is the reason the LT1304 works well with the pairing on the transformer and Cockcraft-Walton multiplier is because of the larger on time (6us), off time (1.5us) which limits the duty cycle to 76%.

    I think if I could find an IC with larger on and off times like the LT1304, I could get the circuit working.  Any help is greatly appreciated.

  • Hi,

    The LT1304 is a different type of switcher, yes, but it will not work for you due to its limited input range, as you have noted.

    Send pictures of your test results, so I can see what you mean, regarding the 'current spikes'.

    DV

  • This simulation is using the LT3580 designed using the suggestion schematic.  The current spikes on the inductor as the voltage ramps up exceeds 100A!!!  This is just to 350V without the charge pump multiplier.  What am I doing wrong?

  • Another using the LT8361 to create 300V shows 10A spikes on the inductor.  Wouldn't that damage the LT8361 because SW is only rated to 2A?

  • Hi, 

    I have built the LT3580 application, and it works fine. You are correct in that those spikes far exceed what the converter would survive, but what you see in simulation is not necessarily true on the bench. 

    If you can get the simulation to get to steady state, focus on the steady state operation of the switch, current and voltage.

    As I said before, ensure the converter runs in deep DCM.

    As a reference for the transformer, please check the leakage inductance in the TDK transformer used in the LT3580 circuit and make sure the transformer you use is as good or better.

    I can take a look at your simulation and tell you if it has a shot on the bench. Post it if you wish.

    Let me know.

    DV

  • How can I verify the converter is running in deep DCM?

  • This will help.

    Waveforms below:

    Switch voltage (green), and switch current (blue) taken from a Boost converter.

    Top=Discontinuous conduction mode, DCM.

    Bottom=Continuous conduction mode, CCM.

  • Thanks for the extra explanation.  How do I make sure the IC operates and stays in DCM mode?  What forces the IC into this mode, low voltage on the Sense pin?

  • DCM operation is mainly dependent on the value of the primary inductance, input voltage level and load current. The IC does not control DCM.

    There are many articles online discussing 'Boost Converter Operation' where you can read details on DCM.

    Your transformer will need 4.7uH inductance (maybe a little more) on the primary in order to keep the switch current within 2A.

    Things you absolutely need to keep in mind:

    1. Keep switch voltage <100V.

    2. Keep switch current <2A.

    3. Keep converter operation in DCM.

    Again, good luck..

    DV